Full Table of Contents¶

Frequently Asked Questions¶

Is Salt open-core?¶

No. Salt is 100% committed to being open-source, including all of our APIs and the new 'Halite' web interface which will be included in version 0.17.0. It is developed under the Apache 2.0 license, allowing it to be used in both open and proprietary projects.

What ports should I open on my firewall?¶

Minions need to be able to connect to the Master on TCP ports 4505 and 4506. Minions do not need any inbound ports open. More detailed information on firewall settings can be found here.

My script runs every time I run a `state.highstate`. Why?¶

You are probably using cmd.run rather than cmd.wait. A cmd.wait state will only run when there has been a change in a state that it is watching.

A cmd.run state will run the corresponding command every time (unless it is prevented from running by the unless or onlyif arguments).

More details can be found in the docmentation for the cmd states.

When I run `test.ping`, why don't the Minions that aren't responding return anything? Returning `False` would be helpful.¶

The reason for this is because the Master tells Minions to run commands/functions, and listens for the return data, printing it to the screen when it is received. If it doesn't receive anything back, it doesn't have anything to display for that Minion.

There are a couple options for getting information on Minions that are not responding. One is to use the verbose (-v) option when you run salt commands, as it will display "Minion did not return" for any Minions which time out.

salt -v '*' pkg.install zsh

Another option is to use the manage.down runner:

salt-run manage.down

How does Salt determine the Minion's id?¶

If the Minion id is not configured explicitly (using the id parameter), Salt will determine the id based on the hostname. Exactly how this is determined varies a little between operating systems and is described in detail here.

I'm using gitfs and my custom modules/states/etc are not syncing. Why?¶

In versions of Salt 0.16.3 or older, there is a bug in gitfs which can affect the syncing of custom types. Upgrading to 0.16.4 or newer will fix this.

Why aren't my custom modules/states/etc. available on my Minions?¶

Custom modules are only synced to Minions when state.highstate, saltutil.sync_modules, or saltutil.sync_all is run. Similarly, custom states are only synced to Minions when state.highstate, saltutil.sync_states, or saltutil.sync_all is run.

Other custom types (renderers, outputters, etc.) have similar behavior, see the documentation for the saltutil module for more information.

Module `X` isn't available, even though the shell command it uses is installed. Why?¶

This is most likely a PATH issue. Did you custom-compile the software which the module requires? RHEL/CentOS/etc. in particular override the root user's path in /etc/init.d/functions, setting it to /sbin:/usr/sbin:/bin:/usr/bin, making software installed into /usr/local/bin unavailable to Salt when the Minion is started using the initscript. In version 0.18.0, Salt will have a better solution for these sort of PATH-related issues, but recompiling the software to install it into a location within the PATH should resolve the issue in the meantime. Alternatively, you can create a symbolic link within the PATH using a file.symlink state.

/usr/bin/foo:
  file.symlink:
    - target: /usr/local/bin/foo

Introduction to Salt¶

We’re not just talking about NaCl.

The 30 second summary¶

Salt is:

a configuration management system, capable of maintaining remote nodes in defined states (for example, ensuring that specific packages are installed and specific services are running)
a distributed remote execution system used to execute commands and query data on remote nodes, either individually or by arbitrary selection criteria

It was developed in order to bring the best solutions found in the world of remote execution together and make them better, faster, and more malleable. Salt accomplishes this through its ability to handle large loads of information, and not just dozens but hundreds and even thousands of individual servers quickly through a simple and manageable interface.

Simplicity¶

Providing versatility between massive scale deployments and smaller systems may seem daunting, but Salt is very simple to set up and maintain, regardless of the size of the project. The architecture of Salt is designed to work with any number of servers, from a handful of local network systems to international deployments across different datacenters. The topology is a simple server/client model with the needed functionality built into a single set of daemons. While the default configuration will work with little to no modification, Salt can be fine tuned to meet specific needs.

Parallel execution¶

The core functions of Salt:

enable commands to remote systems to be called in parallel rather than serially
use a secure and encrypted protocol
use the smallest and fastest network payloads possible
provide a simple programming interface

Salt also introduces more granular controls to the realm of remote execution, allowing systems to be targeted not just by hostname, but also by system properties.

Building on proven technology¶

Salt takes advantage of a number of technologies and techniques. The networking layer is built with the excellent ZeroMQ networking library, so the Salt daemon includes a viable and transparent AMQ broker. Salt uses public keys for authentication with the master daemon, then uses faster AES encryption for payload communication; authentication and encryption are integral to Salt. Salt takes advantage of communication via msgpack, enabling fast and light network traffic.

Python client interface¶

In order to allow for simple expansion, Salt execution routines can be written as plain Python modules. The data collected from Salt executions can be sent back to the master server, or to any arbitrary program. Salt can be called from a simple Python API, or from the command line, so that Salt can be used to execute one-off commands as well as operate as an integral part of a larger application.

Fast, flexible, scalable¶

The result is a system that can execute commands at high speed on target server groups ranging from one to very many servers. Salt is very fast, easy to set up, amazingly malleable and provides a single remote execution architecture that can manage the diverse requirements of any number of servers. The Salt infrastructure brings together the best of the remote execution world, amplifies its capabilities and expands its range, resulting in a system that is as versatile as it is practical, suitable for any network.

Open¶

Salt is developed under the Apache 2.0 license, and can be used for open and proprietary projects. Please submit your expansions back to the Salt project so that we can all benefit together as Salt grows. Please feel free to sprinkle Salt around your systems and let the deliciousness come forth.

Installation¶

The Salt system setup is amazingly simple, as this is one of the central design goals of Salt.

See also

Installing Salt for development and contributing to the project.

Quick Install¶

Many popular distributions will be able to install the salt minion by executing the bootstrap script:

wget -O - http://bootstrap.saltstack.org | sudo sh

Run the following script to install just the Salt Master:

curl -L http://bootstrap.saltstack.org | sudo sh -s -- -M -N

The script should also make it simple to install a salt master, if desired.

Currently the install script has been tested to work on:

Ubuntu 10.x/11.x/12.x
Debian 6.x
CentOS 6.3
Fedora
Arch
FreeBSD 9.0

See Salt Bootstrap for more information.

Platform-specific installation instructions¶

These guides go into detail how to install salt on a given platform.

Arch Linux¶

Installation¶

Salt is currently available via the Arch User Repository (AUR). There are currently stable and -git packages available.

Stable Release¶

Install Salt stable releases from the Arch Linux AUR as follows:

wget https://aur.archlinux.org/packages/sa/salt/salt.tar.gz
tar xf salt.tar.gz
cd salt/
makepkg -is

A few of Salt's dependencies are currently only found within the AUR, so it is necessary to download and run makepkg -is on these as well. As a reference, Salt currently relies on the following packages which are only available via the AUR:

Note

yaourt

If a tool such as Yaourt is used, the dependencies will be gathered and built automatically.

The command to install salt using the yaourt tool is:

yaourt salt

Tracking develop¶

To install the bleeding edge version of Salt (may include bugs!), use the -git package. Installing the -git package as follows:

wget https://aur.archlinux.org/packages/sa/salt-git/salt-git.tar.gz
tar xf salt-git.tar.gz
cd salt-git/
makepkg -is

See the note above about Salt's dependencies.

Post-installation tasks¶

systemd

Activate the Salt Master and/or Minion via systemctl as follows:

systemctl enable salt-master.service
systemctl enable salt-minion.service

Start the Master

Once you've completed all of these steps you're ready to start your Salt Master. You should be able to start your Salt Master now using the command seen here:

systemctl start salt-master

Now go to the Configuring Salt page.

Debian Installation¶

Currently the latest packages for Debian Old Stable, Stable and Unstable (Squeeze, Wheezy and Sid) are published in our (saltstack.com) debian repository.

Configure Apt¶

Squeeze (Old Stable)¶

For squeeze, you will need to enable the debian backports repository as well as the debian.saltstack.com repository. To do so, add the following to /etc/apt/sources.list or a file in /etc/apt/sources.list.d:

deb http://debian.saltstack.com/debian squeeze-saltstack main
deb http://backports.debian.org/debian-backports squeeze-backports main contrib non-free

Wheezy (Stable)¶

For wheezy, the following line is needed in either /etc/apt/sources.list or a file in /etc/apt/sources.list.d:

deb http://debian.saltstack.com/debian wheezy-saltstack main

Sid (Unstable)¶

For sid, the following line is needed in either /etc/apt/sources.list or a file in /etc/apt/sources.list.d:

deb http://debian.saltstack.com/debian unstable main

Import the repository key.¶

You will need to import the key used for signing.

wget -q -O- "http://debian.saltstack.com/debian-salt-team-joehealy.gpg.key" | apt-key add -

Note

You can optionally verify the key integrity with sha512sum using the public key signature shown here. E.g:

echo "b702969447140d5553e31e9701be13ca11cc0a7ed5fe2b30acb8491567560ee62f834772b5095d735dfcecb2384a5c1a20045f52861c417f50b68dd5ff4660e6  debian-salt-team-joehealy.gpg.key" | sha512sum -c

Update the package database¶

apt-get update

Install packages¶

Install the Salt master, minion, or syndic from the repository with the apt-get command. These examples each install one daemon, but more than one package name may be given at a time:

apt-get install salt-master

apt-get install salt-minion

apt-get install salt-syndic

Post-installation tasks¶

Now, go to the Configuring Salt page.

Notes¶

1. These packages will be backported from the packages intended to be uploaded into debian unstable. This means that the packages will be built for unstable first and then backported over the next day or so.

2. These packages will be tracking the released versions of salt rather than maintaining a stable fixed feature set. If a fixed version is what you desire, then either pinning or manual installation may be more appropriate for you.

3. The version numbering and backporting process should provide clean upgrade paths between debian versions.

If you have any questions regarding these, please email the mailing list or look for joehh on irc.

Fedora¶

Beginning with version 0.9.4, Salt has been available in the primary Fedora repositories and EPEL. It is installable using yum. Fedora will have more up to date versions of Salt than other members of the Red Hat family, which makes it a great place to help improve Salt!

Installation¶

Salt can be installed using yum and is available in the standard Fedora repositories.

Stable Release¶

Salt is packaged separately for the minion and the master. It is necessary only to install the appropriate package for the role the machine will play. Typically, there will be one master and multiple minions.

yum install salt-master
yum install salt-minion

Post-installation tasks¶

Master

To have the Master start automatically at boot time:

systemctl enable salt-master.service

To start the Master:

systemctl start salt-master.service

Minion

To have the Minion start automatically at boot time:

systemctl enable salt-minion.service

To start the Minion:

systemctl start salt-minion.service

Now go to the Configuring Salt page.

FreeBSD¶

Salt was added to the FreeBSD ports tree Dec 26th, 2011 by Christer Edwards <christer.edwards@gmail.com>. It has been tested on FreeBSD 7.4, 8.2, 9.0 and 9.1 releases.

Salt is dependent on the following additional ports. These will be installed as dependencies of the sysutils/py-salt port.

/devel/py-yaml
/devel/py-pyzmq
/devel/py-Jinja2
/devel/py-msgpack
/security/py-pycrypto
/security/py-m2crypto

Installation¶

To install Salt from the FreeBSD ports tree, use the command:

make -C /usr/ports/sysutils/py-salt install clean

Post-installation tasks¶

Master

Copy the sample configuration file:

cp /usr/local/etc/salt/master.sample /usr/local/etc/salt/master

rc.conf

Activate the Salt Master in /etc/rc.conf or /etc/rc.conf.local and add:

+ salt_master_enable="YES"

Start the Master

Start the Salt Master as follows:

service salt_master start

Minion

Copy the sample configuration file:

cp /usr/local/etc/salt/minion.sample /usr/local/etc/salt/minion

rc.conf

Activate the Salt Minion in /etc/rc.conf or /etc/rc.conf.local and add:

+ salt_minion_enable="YES"
+ salt_minion_paths="/bin:/sbin:/usr/bin:/usr/sbin:/usr/local/bin:/usr/local/sbin"

Start the Minion

Start the Salt Minion as follows:

service salt_minion start

Now go to the Configuring Salt page.

Gentoo¶

Salt can be easily installed on Gentoo via Portage:

emerge app-admin/salt

Post-installation tasks¶

Now go to the Configuring Salt page.

OS X¶

Dependency Installation¶

ZeroMQ and swig need to be installed first.

For installs using python installed via homebrew, sudo should be unnecessary:

Using homebrew with: XCode Command Line Tool (XCode: Preferences: Downloads: Command Line Tools: Install) pre-installed:

brew install python
brew install swig
brew install zmq
pip install salt

This should pip install salt and its dependencies, such as:: Jinja2 M2Crypto msgpack-python pycrypto PyYAML pyzmq markupsafe

Whereas when using macports, zmq, swig, and pip may need to be installed this way:

sudo port install py-zmq
sudo port install py27-m2crypto
sudo port install py27-crypto
sudo port install py27-msgpack
sudo port install swig-python
sudo port install py-pip

For installs using the OS X system python, pip install needs to use 'sudo':

sudo pip install salt

Salt-Master Customizations¶

To run salt-master on OS X, the root user maxfiles limit must be increased:

sudo launchctl limit maxfiles 4096 8192

And sudo add this configuration option to the /etc/salt/master file:

max_open_files: 8192

Now the salt-master should run without errors:

sudo /usr/local/share/python/salt-master --log-level=all

Post-installation tasks¶

Now go to the Configuring Salt page.

RHEL / CentOS / Scientific Linux / Amazon Linux / Oracle Linux¶

Beginning with version 0.9.4, Salt has been available in EPEL. It is installable using yum. Salt should work properly with all mainstream derivatives of RHEL, including CentOS, Scientific Linux, Oracle Linux and Amazon Linux. Report any bugs or issues to the salt GitHub project.

Installation¶

Salt and all dependencies have been accepted into the yum repositories for EPEL5 and EPEL6. The latest salt version can be found in epel-testing, while an older but more tested version can be found in regular epel.

Example showing how to install salt from epel-testing:

yum --enablerepo=epel-testing install salt-minion

On RHEL6, the proper Jinja package 'python-jinja2' was moved from EPEL to the "RHEL Server Optional Channel". Verify this repository is enabled before installing salt on RHEL6.

Salt can be installed using yum and is available in the standard Fedora repositories.

Enabling EPEL on RHEL¶

If EPEL is not enabled on your system, you can use the following commands to enable it.

For RHEL 5:

rpm -Uvh http://mirror.pnl.gov/epel/5/i386/epel-release-5-4.noarch.rpm

For RHEL 6:

rpm -Uvh http://ftp.linux.ncsu.edu/pub/epel/6/i386/epel-release-6-8.noarch.rpm

Stable Release¶

Salt is packaged separately for the minion and the master. It is necessary only to install the appropriate package for the role the machine will play. Typically, there will be one master and multiple minions.

On the salt-master, run this:

yum install salt-master

On each salt-minion, run this:

yum install salt-minion

Post-installation tasks¶

Master

To have the Master start automatically at boot time:

chkconfig salt-master on

To start the Master:

service salt-master start

Minion

To have the Minion start automatically at boot time:

chkconfig salt-minion on

To start the Minion:

service salt-minion start

Now go to the Configuring Salt page.

Solaris¶

Salt was added to the OpenCSW package repository in September of 2012 by Romeo Theriault <romeot@hawaii.edu> at version 0.10.2 of Salt. It has mainly been tested on Solaris 10 (sparc), though it is built for and has been tested minimally on Solaris 10 (x86), Solaris 9 (sparc/x86) and 11 (sparc/x86). (Please let me know if you're using it on these platforms!) Most of the testing has also just focused on the minion, though it has verified that the master starts up successfully on Solaris 10.

Comments and patches for better support on these platforms is very welcome.

As of version 0.10.4, Solaris is well supported under salt, with all of the following working well:

remote execution
grain detection
service control with SMF
'pkg' states with 'pkgadd' and 'pkgutil' modules
cron modules/states
user and group modules/states
shadow password management modules/states

Salt is dependent on the following additional packages. These will automatically be installed as dependencies of the py_salt package.:

py_yaml
py_pyzmq
py_jinja2
py_msgpack_python
py_m2crypto
py_crypto
python

Installation¶

To install Salt from the OpenCSW package repository you first need to install pkgutil assuming you don't already have it installed:

On Solaris 10:

pkgadd -d http://get.opencsw.org/now

On Solaris 9:

wget http://mirror.opencsw.org/opencsw/pkgutil.pkg
pkgadd -d pkgutil.pkg all

Once pkgutil is installed you'll need to edit it's config file /etc/opt/csw/pkgutil.conf to point it at the unstable catalog:

- #mirror=http://mirror.opencsw.org/opencsw/testing
+ mirror=http://mirror.opencsw.org/opencsw/unstable

OK, time to install salt.

# Update the catalog
root> /opt/csw/bin/pkgutil -U
# Install salt
root> /opt/csw/bin/pkgutil -i -y py_salt

Minion Configuration¶

Now that salt is installed you can find it's configuration files in /etc/opt/csw/salt/.

You'll want to edit the minion config file to set the name of your salt master server:

- #master: salt
+ master: your-salt-server

If you would like to use pkgutil as the default package provider for your Solaris minions, you can do so using the providers option in the minion config file.

You can now start the salt minion like so:

On Solaris 10:

svcadm enable salt-minion

On Solaris 9:

/etc/init.d/salt-minion start

You should now be able to log onto the salt master and check to see if the salt-minion key is awaiting acceptance:

salt-key -l un

Accept the key:

salt-key -a <your-salt-minion>

Run a simple test against the minion:

salt '<your-salt-minion>' test.ping

Troubleshooting¶

Logs are in /var/log/salt

Ubuntu Installation¶

Add repository¶

The latest packages for Ubuntu are published in the saltstack PPA. If you have the add-apt-repository utility, you can add the repository and import the key in one step:

sudo add-apt-repository ppa:saltstack/salt

add-apt-repository: command not found?

The add-apt-repository command is not always present on Ubuntu systems. This can be fixed by installing python-software-properties:

sudo apt-get install python-software-properties

Note that since Ubuntu 12.10 (Raring Ringtail), add-apt-repository is found in the software-properties-common package, and is part of the base install. Thus, add-apt-repository should be able to be used out-of-the-box to add the PPA.

Alternately, manually add the repository and import the PPA key with these commands:

echo deb http://ppa.launchpad.net/saltstack/salt/ubuntu `lsb_release -sc` main | sudo tee /etc/apt/sources.list.d/saltstack.list
wget -q -O- "http://keyserver.ubuntu.com:11371/pks/lookup?op=get&search=0x4759FA960E27C0A6" | sudo apt-key add -

After adding the repository, update the package management database:

sudo apt-get update

Install packages¶

Install the Salt master, minion, or syndic from the repository with the apt-get command. These examples each install one daemon, but more than one package name may be given at a time:

sudo apt-get install salt-master

sudo apt-get install salt-minion

sudo apt-get install salt-syndic

Post-installation tasks¶

Now go to the Configuring Salt page.

Windows¶

Salt has full support for running the Salt Minion on Windows.

There are no plans for the foreseeable future to develop a Salt Master on Windows. For now you must run your Salt Master on a supported operating system to control your Salt Minions on Windows.

Many of the standard Salt modules have been ported to work on Windows and many of the Salt States currently work on Windows, as well.

Windows Installer¶

A Salt Minion Windows installer can be found here:

Download here

The 64bit installer has been tested on Windows 7 64bit and Windows Server 2008R2 64bit. The 32bit installer has been tested on Windows 2003 Server 32bit. Please file a bug report on our GitHub repo if issues for other platforms are found.

The installer asks for 2 bits of information; the master hostname and the minion name. The installer will update the minion config with these options and then start the minion.

The salt-minion service will appear in the Windows Service Manager and can be started and stopped there or with the command line program sc like any other Windows service.

If the minion won't start, try installing the Microsoft Visual C++ 2008 x64 SP1 redistributable. Allow all Windows updates to run salt-minion smoothly.

Make sure that the minion config file has the line ipc_mode: tcp

Silent Installer option¶

The installer can be run silently by providing the /S option at the command line. The options /master and /minion-name allow for configuring the master hostname and minion name, respectively. Here's an example of using the silent installer:

Salt-Minion-0.15.3-Setup-amd64.exe /S /master=yoursaltmaster /minion-name=yourminionname

Installer Source¶

The Salt Windows installer is built with the open-source NSIS compiler. The source for the installer is found in the pkg directory of the Salt repo here: https://github.com/saltstack/salt/blob/develop/pkg/windows/installer/Salt-Minion-Setup.nsi. To create the installer run python setup.py bdist_esky, extract the frozen archive from dist/ into pkg/windows/buildenv/ and run NSIS.

The NSIS installer can be found here: http://nsis.sourceforge.net/Main_Page

Installation from source¶

To install Salt from source one must install each dependency separately and configure Salt to run on your Windows host.

Rather than send you on a wild goose chase across the Internet, we've collected some of the more difficult to find installers in our GitHub repo for you.

Install on Windows XP 32bit¶

Install msysgit
1. Clone the Salt git repository from GitHub

git clone git://github.com/saltstack/salt.git

Install Microsoft Visual Studio 2008 Express. You must use Visual Studio 2008 Express, not Visual Studio 2010 Express.
Install Python 2.7.x
Add c:\Python27 to your system path
Install the Microsoft Visual C++ 2008 SP1 Redistributable, vcredist_x86.
Install Win32OpenSSL-1_0_0e.exe
1. Choose first option to install in Windows system directory
Install pyzmq-2.1.11.win32-py2.7.msi
Install pycrypto-2.3.win32-py2.7.msi
Install M2Crypto
Install pywin32
Install PyYAML-3.10.win32-py2.7.msi
Install Cython-0.15.1.win32-py2.79.exe
Download and run distribute_setup.py

python distribute_setup.py

Download and run pip

python get-pip.py

Add c:\python27\scripts to your path
Close terminal window and open a new terminal window (cmd)
Install jinja2

pip install jinja2

Install wmi

pip install wmi

Install Messagepack

pip install msgpack-python

Install Salt

cd ./salt
python setup.py install

Edit c:\etc\salt\minion

master: ipaddress or hostname of your salt-master
master_port: 4506
ipc_mode: tcp
root_dir: c:\
pki_dir: /etc/salt/pki
cachedir: /var/cache/salt
renderer: yaml_jinja
open_mode: False
multiprocessing: False

Start the salt-minion

cd c:\python27\scripts
python salt-minion

On the salt-master accept the new minion's key

sudo salt-key -A

(This accepts all unaccepted keys. If you're concerned about security just accept the key for this specific minion)

Test that your minion is responding
1. On the salt-master run:

sudo salt '*' test.ping

You should get the following response: {'your minion hostname': True}

Single command bootstrap script¶

On a 64 bit Windows host the following script makes an unattended install of salt, including all dependencies:

Not up to date.

This script is not up to date. Please use the installer found above

"PowerShell (New-Object System.Net.WebClient).DownloadFile('http://csa-net.dk/salt/bootstrap64.bat','C:\bootstrap.bat');(New-Object -com Shell.Application).ShellExecute('C:\bootstrap.bat');"

(All in one line.)

You can execute the above command remotely from a Linux host using winexe:

winexe -U "administrator" //fqdn "PowerShell (New-Object ......);"

For more info check http://csa-net.dk/salt

Packages management under Windows 2003¶

On windows server 2003, you need to install optional component "wmi windows installer provider" to have full list of installed packages. If you don't have this, salt-minion can't report some installed softwares.

SUSE Installation¶

With openSUSE 13.1, Salt 0.16.4 has been available in the primary repositories. The devel:language:python repo will have more up to date versions of salt, all package development will be done there.

Installation¶

Salt can be installed using zypper and is available in the standard openSUSE 13.1 repositories.

Stable Release¶

Salt is packaged separately for the minion and the master. It is necessary only to install the appropriate package for the role the machine will play. Typically, there will be one master and multiple minions.

zypper install salt-master
zypper install salt-minion

Post-installation tasks openSUSE¶

Master

To have the Master start automatically at boot time:

systemctl enable salt-master.service

To start the Master:

systemctl start salt-master.service

Minion

To have the Minion start automatically at boot time:

systemctl enable salt-minion.service

To start the Minion:

systemctl start salt-minion.service

Post-installation tasks SLES¶

Master

To have the Master start automatically at boot time:

chkconfig salt-master on

To start the Master:

rcsalt-master start

Minion

To have the Minion start automatically at boot time:

chkconfig salt-minion on

To start the Minion:

rcsalt-minion start

Unstable Release¶

openSUSE¶

For openSUSE Factory run the following as root:

zypper addrepo http://download.opensuse.org/repositories/devel:languages:python/openSUSE_Factory/devel:languages:python.repo
zypper refresh
zypper install salt salt-minion salt-master

For openSUSE 13.1 run the following as root:

zypper addrepo http://download.opensuse.org/repositories/devel:languages:python/openSUSE_13.1/devel:languages:python.repo
zypper refresh
zypper install salt salt-minion salt-master

For openSUSE 12.3 run the following as root:

zypper addrepo http://download.opensuse.org/repositories/devel:languages:python/openSUSE_12.3/devel:languages:python.repo
zypper refresh
zypper install salt salt-minion salt-master

For openSUSE 12.2 run the following as root:

zypper addrepo http://download.opensuse.org/repositories/devel:languages:python/openSUSE_12.2/devel:languages:python.repo
zypper refresh
zypper install salt salt-minion salt-master

For openSUSE 12.1 run the following as root:

zypper addrepo http://download.opensuse.org/repositories/devel:languages:python/openSUSE_12.1/devel:languages:python.repo
zypper refresh
zypper install salt salt-minion salt-master

For bleeding edge python Factory run the following as root:

zypper addrepo http://download.opensuse.org/repositories/devel:languages:python/bleeding_edge_python_Factory/devel:languages:python.repo
zypper refresh
zypper install salt salt-minion salt-master

Suse Linux Enterprise¶

For SLE 11 SP3 run the following as root:

zypper addrepo http://download.opensuse.org/repositories/devel:languages:python/SLE_11_SP3/devel:languages:python.repo
zypper refresh
zypper install salt salt-minion salt-master

For SLE 11 SP2 run the following as root:

zypper addrepo http://download.opensuse.org/repositories/devel:languages:python/SLE_11_SP2/devel:languages:python.repo
zypper refresh
zypper install salt salt-minion salt-master

Now go to the Configuring Salt page.

Dependencies¶

Salt should run on any Unix-like platform so long as the dependencies are met.

Python 2.6 >= 2.6 <3.0
ZeroMQ >= 2.1.9
pyzmq >= 2.1.9 - ZeroMQ Python bindings
PyCrypto - The Python cryptography toolkit
msgpack-python - High-performance message interchange format
YAML - Python YAML bindings
Jinja2 - parsing Salt States (configurable in the master settings)

Optional Dependencies¶

mako - an optional parser for Salt States (configurable in the master settings)
gcc - dynamic Cython module compiling

Configuring Salt¶

Salt configuration is very simple. The default configuration for the master will work for most installations and the only requirement for setting up a minion is to set the location of the master in the minion configuration file.

master: The Salt master is the central server that all minions connect to. Commands are run on the minions through the master, and minions send data back to the master (unless otherwise redirected with a returner). It is started with the salt-master program.
minion: Salt minions are the potentially hundreds or thousands of servers that may be queried and controlled from the master.

The configuration files will be installed to /etc/salt and are named after the respective components, /etc/salt/master and /etc/salt/minion.

Master Configuration¶

By default the Salt master listens on ports 4505 and 4506 on all interfaces (0.0.0.0). To bind Salt to a specific IP, redefine the "interface" directive in the master configuration file, typically /etc/salt/master, as follows:

- #interface: 0.0.0.0
+ interface: 10.0.0.1

After updating the configuration file, restart the Salt master. See the master configuration reference for more details about other configurable options.

Minion Configuration¶

Although there are many Salt Minion configuration options, configuring a Salt Minion is very simple. By default a Salt Minion will try to connect to the DNS name "salt"; if the Minion is able to resolve that name correctly, no configuration is needed.

If the DNS name "salt" does not resolve to point to the correct location of the Master, redefine the "master" directive in the minion configuration file, typically /etc/salt/minion, as follows:

- #master: salt
+ master: 10.0.0.1

After updating the configuration file, restart the Salt minion. See the minion configuration reference for more details about other configurable options.

Running Salt¶

Start the master in the foreground (to daemonize the process, pass the -d flag):
```
salt-master
```
Start the minion in the foreground (to daemonize the process, pass the -d flag):
```
salt-minion
```

Having trouble?

The simplest way to troubleshoot Salt is to run the master and minion in the foreground with log level set to debug:

salt-master --log-level=debug

For information on salt's logging system please see the logging document.

Run as an unprivileged (non-root) user

To run Salt as another user, specify --user in the command line or assign user in the configuration file.

There is also a full troubleshooting guide available.

Key Management¶

Salt uses AES encryption for all communication between the Master and the Minion. This ensures that the commands sent to the Minions cannot be tampered with, and that communication between Master and Minion is authenticated through trusted, accepted keys.

Before commands can be sent to a Minion, its key must be accepted on the Master. Run the salt-key command to list the keys known to the Salt Master:

[root@master ~]# salt-key -L
Unaccepted Keys:
alpha
bravo
charlie
delta
Accepted Keys:

This example shows that the Salt Master is aware of four Minions, but none of the keys has been accepted. To accept the keys and allow the Minions to be controlled by the Master, again use the salt-key command:

[root@master ~]# salt-key -A
[root@master ~]# salt-key -L
Unaccepted Keys:
Accepted Keys:
alpha
bravo
charlie
delta

The salt-key command allows for signing keys individually or in bulk. The example above, using -A bulk-accepts all pending keys. To accept keys individually use the lowercase of the same option, -a keyname.

See also

salt-key manpage

Sending Commands¶

Communication between the Master and a Minion may be verified by running the test.ping command:

[root@master ~]# salt alpha test.ping
alpha:
    True

Communication between the Master and all Minions may be tested in a similar way:

[root@master ~]# salt '*' test.ping
alpha:
    True
bravo:
    True
charlie:
    True
delta:
    True

Each of the Minions should send a True response as shown above.

What's Next?¶

Understanding targeting is important. From there, depending on the way you wish to use Salt, you should also proceed to learn about States and Execution Modules.

Developing Salt¶

There is a great need for contributions to salt and patches are welcome! The goal here is to make contributions clear, make sure there is a trail for where the code has come from, and most importantly, to give credit where credit is due!

There are a number of ways to contribute to salt development.

Sending a GitHub pull request¶

This is the preferred method for contributions. Simply create a GitHub fork, commit changes to the fork, and then open up a pull request.

The following is an example (from Open Comparison Contributing Docs ) of an efficient workflow for forking, cloning, branching, committing, and sending a pull request for a GitHub repository.

First, make a local clone of your GitHub fork of the salt GitHub repo and make edits and changes locally.

Then, create a new branch on your clone by entering the following commands:

git checkout -b fixed-broken-thing

Switched to a new branch 'fixed-broken-thing'

Choose a name for your branch that describes its purpose.

Now commit your changes to this new branch with the following command:

git commit -am 'description of my fixes for the broken thing'

Note

Using git commit -am, followed by a quoted string, both stages and commits all modified files in a single command. Depending on the nature of your changes, you may wish to stage and commit them separately. Also, note that if you wish to add newly-tracked files as part of your commit, they will not be caught using git commit -am and will need to be added using git add before committing.

Push your locally-committed changes back up to GitHub:

git push --set-upstream origin fixed-broken-thing

Now go look at your fork of the salt repo on the GitHub website. The new branch will now be listed under the "Source" tab where it says "Switch Branches". Select the new branch from this list, and then click the "Pull request" button.

Put in a descriptive comment, and include links to any project issues related to the pull request.

The repo managers will be notified of your pull request and it will be reviewed. If a reviewer asks for changes, just make the changes locally in the same local feature branch, push them to GitHub, then add a comment to the discussion section of the pull request.

Note

Travis-CI

To make reviewing pull requests easier for the maintainers, please enable Travis-CI on your fork. Salt is already configured, so simply follow the first 2 steps on the Travis-CI Getting Started Doc.

Keeping Salt Forks in Sync¶

Salt is advancing quickly. It is therefore critical to pull upstream changes from master into forks on a regular basis. Nothing is worse than putting in a days of hard work into a pull request only to have it rejected because it has diverged too far from master.

To pull in upstream changes:

# For ssh github
git remote add upstream git@github.com:saltstack/salt.git
git fetch upstream

# For https github
git remote add upstream https://github.com/saltstack/salt.git
git fetch upstream

To check the log to be sure that you actually want the changes, run the following before merging:

git log upstream/develop

Then to accept the changes and merge into the current branch:

git merge upstream/develop

For more info, see GitHub Fork a Repo Guide or Open Comparison Contributing Docs

Posting patches to the mailing list¶

Patches will also be accepted by email. Format patches using git format-patch and send them to the Salt users mailing list. The contributor will then get credit for the patch, and the Salt community will have an archive of the patch and a place for discussion.

Installing Salt for development¶

Clone the repository using:

git clone https://github.com/saltstack/salt

Note

Running a self-contained development version¶

During development it is easiest to be able to run the Salt master and minion that are installed in the virtualenv you created above, and also to have all the configuration, log, and cache files contained in the virtualenv as well.

Copy the master and minion config files into your virtualenv:

mkdir -p /path/to/your/virtualenv/etc/salt
cp ./salt/conf/master /path/to/your/virtualenv/etc/salt/master
cp ./salt/conf/minion /path/to/your/virtualenv/etc/salt/minion

Edit the master config file:

Uncomment and change the user: root value to your own user.
Uncomment and change the root_dir: / value to point to /path/to/your/virtualenv.
If you are running version 0.11.1 or older, uncomment and change the pidfile: /var/run/salt-master.pid value to point to /path/to/your/virtualenv/salt-master.pid.
If you are also running a non-development version of Salt you will have to change the publish_port and ret_port values as well.

Edit the minion config file:

Repeat the edits you made in the master config for the user and root_dir values as well as any port changes.
If you are running version 0.11.1 or older, uncomment and change the pidfile: /var/run/salt-minion.pid value to point to /path/to/your/virtualenv/salt-minion.pid.
Uncomment and change the master: salt value to point at localhost.
Uncomment and change the id: value to something descriptive like "saltdev". This isn't strictly necessary but it will serve as a reminder of which Salt installation you are working with.

Note

Using salt-call with a Standalone Minion

If you plan to run salt-call with this self-contained development environment in a masterless setup, you should invoke salt-call with -c /path/to/your/virtualenv/etc/salt so that salt can find the minion config file. Without the -c option, Salt finds its config files in /etc/salt.

Start the master and minion, accept the minion's key, and verify your local Salt installation is working:

cd /path/to/your/virtualenv
salt-master -c ./etc/salt -d
salt-minion -c ./etc/salt -d
salt-key -c ./etc/salt -L
salt-key -c ./etc/salt -A
salt -c ./etc/salt '*' test.ping

Running the master and minion in debug mode can be helpful when developing. To do this, add -l debug to the calls to salt-master and salt-minion. If you would like to log to the console instead of to the log file, remove the -d.

Once the minion starts, you may see an error like the following:

zmq.core.error.ZMQError: ipc path "/path/to/your/virtualenv/var/run/salt/minion/minion_event_7824dcbcfd7a8f6755939af70b96249f_pub.ipc" is longer than 107 characters (sizeof(sockaddr_un.sun_path)).

This means the the path to the socket the minion is using is too long. This is a system limitation, so the only workaround is to reduce the length of this path. This can be done in a couple different ways:

Create your virtualenv in a path that is short enough.
Edit the sock_dir minion config variable and reduce its length. Remember that this path is relative to the value you set in root_dir.

NOTE: The socket path is limited to 107 characters on Solaris and Linux, and 103 characters on BSD-based systems.

Note

File descriptor limits

Ensure that the system open file limit is raised to at least 2047:

# check your current limit
ulimit -n

# raise the limit. persists only until reboot
# use 'limit descriptors 2047' for c-shell
ulimit -n 2047

To set file descriptors on OSX, refer to the OS X Installation instructions.

Using easy_install to Install Salt¶

If you are installing using easy_install, you will need to define a USE_SETUPTOOLS environment variable, otherwise dependencies will not be installed:

USE_SETUPTOOLS=1 easy_install salt

Running the tests¶

You will need mock to run the tests:

pip install mock

If you are on Python < 2.7 then you will also need unittest2:

pip install unittest2

Finally you use setup.py to run the tests with the following command:

./setup.py test

For greater control while running the tests, please try:

./tests/runtests.py -h

Editing and previewing the documentation¶

You need sphinx-build command to build the docs. In Debian/Ubuntu this is provided in the python-sphinx package. Sphinx can also be installed to a virtualenv using pip:

pip install Sphinx

Change to salt documentation directory, then:

cd doc; make html

This will build the HTML docs. Run make without any arguments to see the available make targets, which include html, man, and text.
The docs then are built within the docs/_build/ folder. To update the docs after making changes, run make again.
The docs use reStructuredText for markup. See a live demo at http://rst.ninjs.org/.
The help information on each module or state is culled from the python code that runs for that piece. Find them in salt/modules/ or salt/states/.
To build the docs on Arch Linux, the python2-sphinx package is required. Additionally, it is necessary to tell make where to find the proper sphinx-build binary, like so:

make SPHINXBUILD=sphinx-build2 html

To build the docs on RHEL/CentOS 6, the python-sphinx10 package must be installed from EPEL, and the following make command must be used:

make SPHINXBUILD=sphinx-1.0-build html

Targeting¶

Targeting: Specifying which minions should run a command or execute a state by matching against hostnames, or system information, or defined groups, or even combinations thereof.

For example the command salt web1 apache.signal restart to restart the Apache httpd server specifies the machine web1 as the target and the command will only be run on that one minion.

Similarly when using States, the following top file specifies that only the web1 minion should execute the contents of webserver.sls:

base:
  'web1':
    - webserver

There are many ways to target individual minions or groups of minions in Salt:

Matching the `minion id`¶

minion id: A unique identifier for a given minion. By default the minion id is the FQDN of that host but this can be overridden.

Each minion needs a unique identifier. By default when a minion starts for the first time it chooses its FQDN as that identifier. The minion id can be overridden via the minion's id configuration setting.

Tip

minion id and minion keys

The minion id is used to generate the minion's public/private keys and if it ever changes the master must then accept the new key as though the minion was a new host.

Globbing¶

The default matching that Salt utilizes is shell-style globbing around the minion id. This also works for states in the top file.

Note

You must wrap salt calls that use globbing in single-quotes to prevent the shell from expanding the globs before Salt is invoked.

Match all minions:

salt '*' test.ping

Match all minions in the example.net domain or any of the example domains:

salt '*.example.net' test.ping
salt '*.example.*' test.ping

Match all the webN minions in the example.net domain (web1.example.net, web2.example.net … webN.example.net):

salt 'web?.example.net' test.ping

Match the web1 through web5 minions:

salt 'web[1-5]' test.ping

Match the web-x, web-y, and web-z minions:

salt 'web-[x-z]' test.ping

Regular Expressions¶

Minions can be matched using Perl-compatible regular expressions (which is globbing on steroids and a ton of caffeine).

Match both web1-prod and web1-devel minions:

salt -E 'web1-(prod|devel)' test.ping

When using regular expressions in a State's top file, you must specify the matcher as the first option. The following example executes the contents of webserver.sls on the above-mentioned minions.

base:
  'web1-(prod|devel)':
  - match: pcre
  - webserver

Lists¶

At the most basic level, you can specify a flat list of minion IDs:

salt -L 'web1,web2,web3' test.ping

Grains¶

Salt comes with an interface to derive information about the underlying system. This is called the grains interface, because it presents salt with grains of information.

Grains: Static bits of information that a minion collects about the system when the minion first starts.

The grains interface is made available to Salt modules and components so that the right salt minion commands are automatically available on the right systems.

It is important to remember that grains are bits of information loaded when the salt minion starts, so this information is static. This means that the information in grains is unchanging, therefore the nature of the data is static. So grains information are things like the running kernel, or the operating system.

Match all CentOS minions:

salt -G 'os:CentOS' test.ping

Match all minions with 64-bit CPUs, and return number of CPU cores for each matching minion:

salt -G 'cpuarch:x86_64' grains.item num_cpus

Additionally, globs can be used in grain matches, and grains that are nested in a dictionary can be matched by adding a colon for each level that is traversed. For example, the following will match hosts that have a grain called ec2_tags, which itself is a dict with a key named environment, which has a value that contains the word production:

salt -G 'ec2_tags:environment:*production*'

Listing Grains¶

Available grains can be listed by using the 'grains.ls' module:

salt '*' grains.ls

Grains data can be listed by using the 'grains.items' module:

salt '*' grains.items

Grains in the Minion Config¶

Grains can also be statically assigned within the minion configuration file. Just add the option grains and pass options to it:

grains:
  roles:
    - webserver
    - memcache
  deployment: datacenter4
  cabinet: 13
  cab_u: 14-15

Then status data specific to your servers can be retrieved via Salt, or used inside of the State system for matching. It also makes targeting, in the case of the example above, simply based on specific data about your deployment.

Grains in /etc/salt/grains¶

If you do not want to place your custom static grains in the minion config file, you can also put them in /etc/salt/grains. They are configured in the same way as in the above example, only without a top-level grains: key:

roles:
  - webserver
  - memcache
deployment: datacenter4
cabinet: 13
cab_u: 14-15

Precedece of Custom Static Grains

Be careful when defining grains both in /etc/salt/grains and within the minion config file. If a grain is defined in both places, the value in the minion config file takes precedence, and will always be used over its counterpart in /etc/salt/grains.

Grains in Top file¶

With correctly setup grains on the Minion, the Top file used in Pillar or during Highstate can be made really efficient. Like for example, you could do:

'node_type:web':
    - match: grain
    - webserver

'node_type:postgres':
    - match: grain
    - database

'node_type:redis':
    - match: grain
    - redis

'node_type:lb':
    - match: grain
    - lb

For this example to work, you would need the grain node_type and the correct value to match on. This simple example is nice, but too much of the code is similar. To go one step further, we can place some Jinja template code into the Top file.

{% set self = grains['node_type'] %}

    'node_type:{{ self }}':
        - match: grain
        - {{ self }}

The Jinja code simplified the Top file, and allowed SaltStack to work its magic.

Writing Grains¶

Grains are easy to write. The grains interface is derived by executing all of the "public" functions found in the modules located in the grains package or the custom grains directory. The functions in the modules of the grains must return a Python dict, where the keys in the dict are the names of the grains and the values are the values.

Custom grains should be placed in a _grains directory located under the file_roots specified by the master config file. They will be distributed to the minions when state.highstate is run, or by executing the saltutil.sync_grains or saltutil.sync_all functions.

Before adding a grain to Salt, consider what the grain is and remember that grains need to be static data. If the data is something that is likely to change, consider using Pillar instead.

Examples of Grains¶

The core module in the grains package is where the main grains are loaded by the Salt minion and provides the principal example of how to write grains:

https://github.com/saltstack/salt/blob/develop/salt/grains/core.py

Syncing Grains¶

Syncing grains can be done a number of ways, they are automatically synced when state.highstate is called, or (as noted above) the grains can be manually synced and reloaded by calling the saltutil.sync_grains or saltutil.sync_all functions.

Node groups¶

Node group: A predefined group of minions declared in the master configuration file nodegroups setting as a compound target.

Nodegroups are declared using a compound target specification. The compound target documentation can be found here.

The nodegroups master config file parameter is used to define nodegroups. Here's an example nodegroup configuration:

nodegroups:
  group1: 'L@foo.domain.com,bar.domain.com,baz.domain.com or bl*.domain.com'
  group2: 'G@os:Debian and foo.domain.com'

To match a nodegroup on the CLI, use the -N command-line option:

salt -N group1 test.ping

To match in your top file, make sure to put - match: nodegroup on the line directly following the nodegroup name.

base:
  group1:
    - match: nodegroup
    - webserver

Compound matchers¶

Compound matcher: A combination of many target definitions that can be combined with boolean operators.

Compound matchers allow very granular minion targeting using any of Salt's matchers. The default matcher is a glob match, just as with CLI and top file matching. To match using anything other than a glob, prefix the match string with the appropriate letter from the table below, followed by an @ sign.

Letter	Match Type	Example
G	Grains glob	`G@os:Ubuntu`
E	PCRE Minion ID	`E@web\d+\.(dev\|qa\|prod)\.loc`
P	Grains PCRE	`P@os:(RedHat\|Fedora\|CentOS)`
L	List of minions	`L@minion1.example.com,minion3.domain.com or bl*.domain.com`
I	Pillar glob	`I@pdata:foobar`
S	Subnet/IP address	`S@192.168.1.0/24` or `S@192.168.1.100`
R	Range cluster	`R@%foo.bar`

Matchers can be joined using boolean and, or, and not operators.

For example, the following string matches all Debian minions with a hostname that begins with webserv, as well as any minions that have a hostname which matches the regular expression web-dc1-srv.*:

salt -C 'webserv* and G@os:Debian or E@web-dc1-srv.*' test.ping

That same example expressed in a top file looks like the following:

base:
  'webserv* and G@os:Debian or E@web-dc1-srv.*':
    - match: compound
    - webserver

Note that a leading not is not supported in compound matches. Instead, something like the following must be done:

salt -C '* and not G@kernel:Darwin' test.ping

Batch Size¶

The -b (or --batch-size) option allows commands to be executed on only a specified number of minions at a time. Both percentages and finite numbers are supported.

salt '*' -b 10 test.ping

salt -G 'os:RedHat' --batch-size 25% apache.signal restart

This will only run test.ping on 10 of the targeted minions at a time and then restart apache on 25% of the minions matching os:RedHat at a time and work through them all until the task is complete. This makes jobs like rolling web server restarts behind a load balancer or doing maintenance on BSD firewalls using carp much easier with salt.

The batch system maintains a window of running minions, so, if there are a total of 150 minions targeted and the batch size is 10, then the command is sent to 10 minions, when one minion returns then the command is sent to one additional minion, so that the job is constantly running on 10 minions.

Salt tutorials¶

Bootstrapping Salt on Linux EC2 with Cloud-Init¶

Salt is a great tool for remote execution and configuration management, however you will still need to bootstrap the daemon when spinning up a new node. One option is to create and save a custom AMI, but this creates another resource to maintain and document.

A better method for Linux machines uses Canonical's CloudInit to run a bootstrap script during an EC2 Instance initialization. Cloud-init takes the user_data string passed into a new AWS instance and runs it in a manner similar to rc.local. The bootstrap script needs to:

Install Salt with dependencies
Point the minion to the master

Here is a sample script:

#!/bin/bash

# Install saltstack
add-apt-repository ppa:saltstack/salt -y
apt-get update -y
apt-get install salt-minion -y
apt-get install salt-master -y
apt-get upgrade -y

# Set salt master location and start minion
sed -i 's/#master: salt/master: [salt_master_fqdn]/' /etc/salt/minion
salt-minion -d

First the script adds the saltstack ppa and installs the package. Then we copy over the minion config template and tell it where to find the master. You will have to replace [salt_master_fqdn] with something that resolves to your Salt master.

Used With Boto¶

Boto will accept a string for user data which can be used to pass our bootstrap script. If the script is saved to a file, you can read it into a string:

import boto

user_data = open('salt_bootstrap.sh')

conn = boto.connect_ec2(<AWS_ACCESS_ID>, <AWS_SECRET_KEY>)

reservation = conn.run_instances(image_id=<ami_id>,
                                 key_name=<key_name>,
                                 user_data=user_data.read())

Additional Notes¶

Sometime in the future the ppa will include and install an upstart file. In the meantime, you can use the bootstrap to build one.

It may also be useful to set the node's role during this phase. One option would be saving the node's role to a file and then using a custom Grain to select it.

Salt as a Cloud Controller¶

In Salt 0.14.0 advanced cloud control systems were introduced, allowing for private cloud vms to be managed directly with Salt. This system is generally referred to as Salt Virt.

The Salt Virt system already exists and is installed within Salt itself, this means that beyond setting up Salt no additional salt code needs to be deployed.

Setting up Hypervisors¶

The first step to set up the hypervisors involves getting the correct software installed and setting up the hypervisor network interfaces.

Installing Hypervisor Software¶

Salt Virt is made to be hypervisor agnostic, but currently the only implemented hypervisor is KVM via libvirt.

The required software for a hypervisor is libvirt and kvm. For advanced features install libguestfs or qemu-nbd.

Note

Libguestfs and qemu-nbd allow for virtual machine images to be mounted before startup and get pre-seeded with configurations and a salt minion

A simple sls formula to deploy the required software and service:

Note

Package names used are Red Hat specific, different package names will be required for different platforms

libguestfs:
  pkg.installed

qemu-kvm:
  pkg.installed

libvirt:
  pkg.installed

libvirtd:
  service.running:
    - enable: True
    - watch:
      - pkg: libvirt

Network Setup¶

Salt virt comes with a system to model the network interfaces used by the deployed virtual machines, by default a single interface is created for the deployed virtual machine and is bridged to br0. To get going with the default networking setup ensure that the bridge interface named br0 exists on the hypervisor and is bridged to an active network device.

Note

To use more advanced networking in Salt Virt read the Salt Virt Networking document:

Salt Virt Networking

Libvirt State¶

One of the challenges of deploying a libvirt based cloud is the distribution of libvirt certificates. These certificates allow for virtual machine migration. Salt comes with a system used to auto deploy these certificates. Salt manages the signing authority key and generates keys for libvirt clients on the master, signs them with the certificate authority and uses pillar to distribute them. This is managed via the libvirt state. Simply execute this formula on the minion to ensure that the certificate is in place and up to date:

libvirt_keys:
  libvirt.keys

Getting Virtual Machine Images Ready¶

Salt Virt, requires that virtual machine images be provided as these are not generated on the fly. Generating these virtual machine images differs greatly based on the underlying platform.

Virtual machine images can be manually created using KVM and running through the installer, but this process is not recommended since it is very manual and prone to errors.

Virtual Machine generation applications are avilable for many platforms:

vm-builder:: http://wiki.debian.org/VMBuilder

Using Salt Virt¶

With hypervisors set up and virtual machine images ready, Salt can start issuing cloud commands.

Start by deploying

Using cron with Salt¶

The Salt Minion can initiate its own highstate using the salt-call command.

$ salt-call state.highstate

This will cause the minion to check in with the master and ensure it is in the correct 'state'.

Use cron to initiate a highstate¶

If you would like the Salt Minion to regularly check in with the master you can use the venerable cron to run the salt-call command.

# PATH=/bin:/sbin:/usr/bin:/usr/sbin

00 00 * * * salt-call state.highstate

The above cron entry will run a highstate every day at midnight.

Note

Be aware that you may need to ensure the PATH for cron includes any scripts or commands that need to be executed.

Automatic Updates / Frozen Deployments¶

New in version 0.10.3.d.

Salt has support for the Esky application freezing and update tool. This tool allows one to build a complete zipfile out of the salt scripts and all their dependencies - including shared objects / DLLs.

Getting Started¶

To build frozen applications, you'll need a suitable build environment for each of your platforms. You should probably set up a virtualenv in order to limit the scope of Q/A.

This process does work on Windows. Follow the directions at https://github.com/saltstack/salt-windows-install for details on installing Salt in Windows. Only the 32-bit Python and dependencies have been tested, but they have been tested on 64-bit Windows.

You will need to install esky and bbfreeze from Pypi in order to enable the bdist_esky command in setup.py.

Building and Freezing¶

Once you have your tools installed and the environment configured, you can then python setup.py bdist to get the eggs prepared. After that is done, run python setup.py bdist_esky to have Esky traverse the module tree and pack all the scripts up into a redistributable. There will be an appropriately versioned salt-VERSION.zip in dist/ if everything went smoothly.

Windows¶

You will need to add C:\Python27\lib\site-packages\zmq to your PATH variable. This helps bbfreeze find the zmq dll so it can pack it up.

Using the Frozen Build¶

Unpack the zip file in your desired install location. Scripts like salt-minion and salt-call will be in the root of the zip file. The associated libraries and bootstrapping will be in the directories at the same level. (Check the Esky documentation for more information)

To support updating your minions in the wild, put your builds on a web server that your minions can reach. salt.modules.saltutil.update() will trigger an update and (optionally) a restart of the minion service under the new version.

Gotchas¶

My Windows minion isn't responding¶

The process dispatch on Windows is slower than it is on *nix. You may need to add '-t 15' to your salt calls to give them plenty of time to return.

Windows and the Visual Studio Redist¶

You will need to install the Visual C++ 2008 32-bit redistributable on all Windows minions. Esky has an option to pack the library into the zipfile, but OpenSSL does not seem to acknowledge the new location. If you get a no OPENSSL_Applink error on the console when trying to start your frozen minion, you have forgotten to install the redistributable.

Mixed Linux environments and Yum¶

The Yum Python module doesn't appear to be available on any of the standard Python package mirrors. If you need to support RHEL/CentOS systems, you should build on that platform to support all your Linux nodes. Also remember to build your virtualenv with --system-site-packages so that the yum module is included.

Automatic (Python) module discovery¶

Automatic (Python) module discovery does not work with the late-loaded scheme that Salt uses for (Salt) modules. You will need to explicitly add any misbehaving modules to the freezer_includes in Salt's setup.py. Always check the zipped application to make sure that the necessary modules were included.

Opening the Firewall up for Salt¶

The Salt master communicates with the minions using an AES-encrypted ZeroMQ connection. These communications are done over TCP ports 4505 and 4506, which need to be accessible on the master only. This document outlines suggested firewall rules for allowing these incoming connections to the master.

Note

No firewall configuration needs to be done on Salt minions. These changes refer to the master only.

RHEL 6 / CENTOS 6¶

The lokkit command packaged with some Linux distributions makes opening iptables firewall ports very simple via the command line. Just be careful to not lock out access to the server by neglecting to open the ssh port.

lokkit example:

lokkit -p 22:tcp -p 4505:tcp -p 4506:tcp

The system-config-firewall-tui command provides a text-based interface to modifying the firewall.

system-config-firewall-tui:

system-config-firewall-tui

openSUSE¶

Salt installs firewall rules in /etc/sysconfig/SuSEfirewall2.d/services/salt. Enable with:

SuSEfirewall2 open
SuSEfirewall2 start

If you have an older package of Salt where the above configuration file is not included, the SuSEfirewall2 command makes opening iptables firewall ports very simple via the command line.

SuSEfirewall example:

SuSEfirewall2 open EXT TCP 4505
SuSEfirewall2 open EXT TCP 4506

The firewall module in YaST2 provides a text-based interface to modifying the firewall.

YaST2:

yast2 firewall

iptables¶

Different Linux distributions store their iptables rules in different places, which makes it difficult to standardize firewall documentation. Included are some of the more common locations, but your mileage may vary.

Fedora / RHEL / CentOS:

/etc/sysconfig/iptables

Arch Linux:

/etc/iptables/iptables.rules

Debian

Follow these instructions: http://wiki.debian.org/iptables

Once you've found your firewall rules, you'll need to add the two lines below to allow traffic on tcp/4505 and tcp/4506:

-A INPUT -m state --state new -m tcp -p tcp --dport 4505 -j ACCEPT
-A INPUT -m state --state new -m tcp -p tcp --dport 4506 -j ACCEPT

Ubuntu

Salt installs firewall rules in /etc/ufw/applications.d/salt.ufw. Enable with:

ufw allow salt

pf.conf¶

The BSD-family of operating systems uses packet filter (pf). The following example describes the additions to pf.conf needed to access the Salt master.

pass in on $int_if proto tcp from any to $int_if port 4505
pass in on $int_if proto tcp from any to $int_if port 4506

Once these additions have been made to the pf.conf the rules will need to be reloaded. This can be done using the pfctl command.

pfctl -vf /etc/pf.conf

GitFS Backend Walkthrough¶

While the default location of the salt state tree is on the Salt master, in /srv/salt, the master can create a bridge to external resources for files. One of these resources is the ability for the master to directly pull files from a git repository and serve them to minions.

Note

This walkthrough assumes basic knowledge of Salt. To get up to speed, check out the walkthrough.

The gitfs backend hooks into any number of remote git repositories and caches the data from the repository on the master. This makes distributing a state tree to multiple masters seamless and automated.

Salt's file server also has a concept of environments, when using the gitfs backend, Salt translates git branches and tags into environments, making environment management very simple. Just merging a QA or staging branch up to a production branch can be all that is required to make those file changes available to Salt.

Simple Configuration¶

To use the gitfs backend only two configuration changes are required on the master. The fileserver_backend option needs to be set with a value of git:

fileserver_backend:
  - git

To configure what fileserver backends will be searched for requested files.

Now the gitfs system needs to be configured with a remote:

gitfs_remotes:
  - git://github.com/saltstack/salt-states.git

Note

The salt-states repo is not currently updated with the latest versions of the available states. Please review https://github.com/saltstack-formulas for the latest versions.

These changes require a restart of the master, then the git repo will be cached on the master and new requests for the salt:// protocol will send files found in the remote git repository via the master.

Note

The master caches the files from the git server and serves them out, minions do not connect directly to the git server meaning that only requested files are delivered to minions.

Multiple Remotes¶

The gitfs_remotes option can accept a list of git remotes, the remotes are then searched in order for the requested file. A simple scenario can illustrate this behavior.

Assuming that the gitfs_remotes option specifies three remotes:

gitfs_remotes:
  - git://github.com/example/first.git
  - git://github.com/example/second.git
  - file:///root/third

Note

This example is purposefully contrived to illustrate the behavior of the gitfs backend. This example should not be read as a recommended way to lay out files and git repos.

Note

The file:// prefix denotes a git repository in a local directory. However, it will still use the given file:// URL as a remote, rather than copying the git repo to the salt cache. This means that any refs you want accessible must exist as local refs in the specified repo.

Assume that each repository contains some files:

first.git:
    top.sls
    edit/vim.sls
    edit/vimrc
    nginx/init.sls

second.git:
    edit/dev_vimrc
    haproxy/init.sls

third:
    haproxy/haproxy.conf
    edit/dev_vimrc

The repositories will be searched for files by the master in the order in which they are defined in the configuration, Therefore the remote git://github.com/example/first.git will be searched first, if the requested file is found then it is served and no further searching is executed. This means that if the file salt://haproxy/init.sls is requested then it will be pulled from the git://github.com/example/second.git git repo. If salt://haproxy/haproxy.conf is requested then it will be pulled from the third repo.

Serving from a Subdirectory¶

The gitfs_root option gives the ability to serve files from a subdirectory within the repository. The path is defined relative to the root of the repository.

With this repository structure:

repository.git:
    somefolder
        otherfolder
            top.sls
            edit/vim.sls
            edit/vimrc
            nginx/init.sls

Configuration and files can be accessed normally with:

gitfs_root: somefolder/otherfolder

Multiple Backends¶

Sometimes it may make sense to use multiple backends. For instance, if sls files are stored in git, but larger files need to be stored directly on the master.

The logic used for multiple remotes is also used for multiple backends. If the fileserver_backend option contains multiple backends:

fileserver_backend:
  - roots
  - git

Then the roots backend (the default backend of files in /srv/salt) will be searched first for the requested file, then if it is not found on the master the git remotes will be searched.

Branches, environments and top.sls files¶

As stated above, when using the gitfs backend, branches will be mapped to environments using the branch name as identifier. There is an exception to this rule thought: the master branch is implicitly mapped to the base environment. Therefore, for a typical base, qa, dev setup, you'll have to create the following branches:

master
qa
dev

Also, top.sls files from different branches will be merged into one big file at runtime. Since this could lead to hardly manageable configurations, the recommended setup is to have the top.sls file only in your master branch, and use environment-specific branches for states definitions.

GitFS Remotes over SSH¶

In order to configure a gitfs_remotes repository over SSH transport the git+ssh URL form must be used.

gitfs_remotes:
  - git+ssh://git@github.com/example/salt-states.git

The private key used to connect to the repository must be located in ~/.ssh/id_rsa for the user running the salt-master.

Note

GitFS requires the Python module GitPython, version 0.3.0 or newer.

Why aren't my custom modules/states/etc. syncing to my Minions?¶

In versions 0.16.3 and older, when using the git fileserver backend, certain versions of GitPython may generate errors when fetching, which Salt fails to catch. While not fatal to the fetch process, these interrupt the fileserver update that takes place before custom types are synced, and thus interrupt the sync itself. Try disabling the git fileserver backend in the master config, restarting the master, and attempting the sync again.

This issue will be worked around in Salt 0.16.4 and newer.

Remote execution tutorial¶

Before continuing make sure you have a working Salt installation by following the installation and the configuration instructions.

Stuck?

There are many ways to get help from the Salt community including our mailing list and our IRC channel #salt.

Order your minions around¶

Now that you have a master and at least one minion communicating with each other you can perform commands on the minion via the salt command. Salt calls are comprised of three main components:

salt '<target>' <function> [arguments]

See also

salt manpage

target¶

The target component allows you to filter which minions should run the following function. The default filter is a glob on the minion id. For example:

salt '*' test.ping
salt '*.example.org' test.ping

Targets can be based on minion system information using the Grains system:

salt -G 'os:Ubuntu' test.ping

See also

Grains system

Targets can be filtered by regular expression:

salt -E 'virtmach[0-9]' test.ping

Targets can be explicitly specified in a list:

salt -L 'foo,bar,baz,quo' test.ping

Or Multiple target types can be combined in one command:

salt -C 'G@os:Ubuntu and webser* or E@database.*' test.ping

function¶

A function is some functionality provided by a module. Salt ships with a large collection of available functions. List all available functions on your minions:

salt '*' sys.doc

Here are some examples:

Show all currently available minions:

salt '*' test.ping

Run an arbitrary shell command:

salt '*' cmd.run 'uname -a'

See also

the full list of modules

arguments¶

Space-delimited arguments to the function:

salt '*' cmd.exec_code python 'import sys; print sys.version'

Optional, keyword arguments are also supported:

salt '*' pip.install salt timeout=5 upgrade=True

They are always in the form of kwarg=argument.

Multi Master Tutorial¶

As of Salt 0.16.0, the ability to connect minions to multiple masters has been made available. The multi-master system allows for redundancy of Salt masters and facilitates multiple points of communication out to minions. When using a multi-master setup, all masters are running hot, and any active master can be used to send commands out to the minions.

In 0.16.0, the masters do not share any information, keys need to be accepted on both masters, and shared files need to be shared manually or use tools like the git fileserver backend to ensure that the file_roots are kept consistent.

Summary of Steps¶

Create a redundant master server
Copy primary master key to redundant master
Start redundant master
Configure minions to connect to redundant master
Restart minions
Accept keys on redundant master

Prepping a Redundant Master¶

The first task is to prepare the redundant master. There is only one requirement when preparing a redundant master, which is that masters share the same private key. When the first master was created, the master's identifying key was generated and placed in the master's pki_dir. The default location of the key is /etc/salt/pki/master/master.pem. Take this key and copy it to the same location on the redundant master. Assuming that no minions have yet been connected to the new redundant master, it is safe to delete any existing key in this location and replace it.

Note

There is no logical limit to the number of redundant masters that can be used.

Once the new key is in place, the redundant master can be safely started.

Configure Minions¶

Since minions need to be master-aware, the new master needs to be added to the minion configurations. Simply update the minion configurations to list all connected masters:

master:
  - saltmaster1.example.com
  - saltmaster2.example.com

Now the minion can be safely restarted.

Now the minions will check into the original master and also check into the new redundant master. Both masters are first-class and have rights to the minions.

Pillar Walkthrough¶

Note

This walkthrough assumes that the reader has already completed the initial Salt Stack walkthrough.

The pillar interface inside of Salt is one of the most important components of a Salt deployment. Pillar is the interface used to generate arbitrary data for specific minions. The data generated in pillar is made available to almost every component of Salt and is used for a number of purposes:

Highly Sensitive Data:: Information transferred via pillar is guaranteed to only be presented to the minions that are targeted, this makes pillar the engine to use in Salt for managing security information, such as cryptographic keys and passwords.
Minion Configuration:: Minion modules such as the execution modules, states, and returners can often be configured via data stored in pillar.
Variables:: Variables which need to be assigned to specific minions or groups of minions can be defined in pillar and then accessed inside sls formulas and template files.
Arbitrary Data:: Pillar can contain any basic data structure, so a list of values, or a key/value store can be defined making it easy to iterate over a group of values in sls formulas

Pillar is therefore one of the most important systems when using Salt, this walkthrough is designed to get a simple pillar up and running in a few minutes and then to dive into the capabilities of pillar and where the data is available.

Setting Up Pillar¶

The pillar is already running in Salt by default. The data in the minion's pillars can be seen via the following command:

salt '*' pillar.items

Note

Prior to version 0.16.2, this function is named pillar.data. This function name is still supported for backwards compatibility.

By default the contents of the master configuration file are loaded into pillar for all minions, this is to enable the master configuration file to be used for global configuration of minions.

The pillar is built in a similar fashion as the state tree, it is comprised of sls files and has a top file, just like the state tree. The pillar is stored in a different location on the Salt master than the state tree. The default location for the pillar is in /srv/pillar.

Note

The pillar location can be configured via the pillar_roots option inside the master configuration file.

To start setting up the pillar, the /srv/pillar directory needs to be present:

mkdir /srv/pillar

Now a simple top file, following the same format as the top file used for states needs to be created:

/srv/pillar/top.sls:

base:
  '*':
    - data

This top file associates the data.sls file to all minions. Now the /srv/pillar/data.sls file needs to be populated:

/srv/pillar/data.sls:

info: some data

Now that the file has been saved the minions' pillars will be updated:

salt '*' pillar.items

The key info should now appear in the returned pillar data.

More Complex Data¶

Pillar files are sls files, just like states, but unlike states they do not need to define formulas, the data can be arbitrary, this example for instance sets up user data with a UID:

/srv/pillar/users/init.sls:

users:
  thatch: 1000
  shouse: 1001
  utahdave: 1002
  redbeard: 1003

Note

The same directory lookups that exist in states exist in pillar, so the file users/init.sls can be referenced with users in the top file.

The top file will need to be updated to include this sls file:

/srv/pillar/top.sls:

base:
  '*':
    - data
    - users

Now the data will be available to the minions. To use the pillar data in a state just access the pillar via Jinja:

/srv/salt/users/init.sls

{% for user, uid in pillar.get('users', {}).items() %}
{{user}}:
  user.present:
    - uid: {{uid}}
{% endfor %}

This approach allows for users to be safely defined in a pillar and then the user data is applied in an sls file.

Paramaterizing States With Pillar¶

One of the most powerful abstractions in pillar is the ability to parameterize states. Instead of defining macros or functions within the state context the entire state tree can be freely parameterized relative to the minion's pillar.

This approach allows for Salt to be very flexible while staying very straightforward. It also means that simple sls formulas used in the state tree can be directly parameterized without needing to refactor the state tree.

A simple example is to set up a mapping of package names in pillar for separate Linux distributions:

/srv/pillar/pkg/init.sls:

pkgs:
  {% if grains['os_family'] == 'RedHat' %}
  apache: httpd
  vim: vim-enhanced
  {% elif grains['os_family'] == 'Debian' %}
  apache: apache2
  vim: vim
  {% elif grains['os'] == 'Arch' %}
  apache: apache
  vim: vim
  {% endif %}

The new pkg sls needs to be added to the top file:

/srv/pillar/top.sls:

base:
  '*':
    - data
    - users
    - pkg

Now the minions will auto map values based on respective operating systems inside of the pillar, so sls files can be safely parameterized:

/srv/salt/apache/init.sls:

apache:
  pkg.installed:
    - name: {{ pillar['pkgs']['apache'] }}

Or, if no pillar is available a default can be set as well:

Note

The function pillar.get used in this example was added to Salt in version 0.14.0

/srv/salt/apache/init.sls:

apache:
  pkg.installed:
    - name: {{ salt['pillar.get']('pkgs:apache', 'httpd') }}

In the above example, if the pillar value pillar['pkgs']['apache'] is not set in the minion's pillar, then the default of httpd will be used.

Note

Under the hood, pillar is just a python dict, so python dict methods such as get and items can be used.

Pillar Makes Simple States Grow Easily¶

One of the design goals of pillar is to make simple sls formulas easily grow into more flexible formulas without refactoring or complicating the states.

A simple formula:

/srv/salt/edit/vim.sls:

vim:
  pkg:
    - installed

/etc/vimrc:
  file.managed:
    - source: salt://edit/vimrc
    - mode: 644
    - user: root
    - group: root
    - require:
      - pkg: vim

Can be easily transformed into a powerful, parameterized formula:

/srv/salt/edit/vim.sls:

vim:
  pkg:
    - installed
    - name: {{ pillar['pkgs']['vim'] }}

/etc/vimrc:
  file.managed:
    - source: {{ pillar['vimrc'] }}
    - mode: 644
    - user: root
    - group: root
    - require:
      - pkg: vim

Where the vimrc source location can now be changed via pillar:

/srv/pillar/edit/vim.sls:

{% if grain['id'].startswith('dev') %}
vimrc: salt://edit/dev_vimrc
{% elif grain['id'].startswith('qa') %}
vimrc: salt://edit/qa_vimrc
{% else %}
vimrc: salt://edit/vimrc
{% endif %}

Ensuring that the right vimrc is sent out to the correct minions.

More On Pillar¶

The pillar data is generated on the Salt master and securely distributed to minions. Salt is not restricted to the pillar sls files when defining the pillar but can retrieve data from external sources. This can be useful when information about an infrastructure is stored in a separate location.

Reference information on pillar and the external pillar interface can be found in the Salt Stack documentation:

Pillar

Preseed Minion with Accepted Key¶

In some situations, it is not convenient to wait for a minion to start before accepting its key on the master. For instance, you may want the minion to bootstrap itself as soon as it comes online. You may also want to to let your developers provision new development machines on the fly.

There is a general four step process to do this:

Generate the keys on the master:

root@saltmaster# salt-key --gen-keys=[key_name]

Pick a name for the key, such as the minion's id.

Add the public key to the accepted minion folder:

root@saltmaster# cp key_name.pub /etc/salt/pki/master/minions/[minion_id]

It is necessary that the public key file has the same name as your minion id. This is how Salt matches minions with their keys. Also note that the pki folder could be in a different location, depending on your OS or if specified in the master config file.

Distribute the minion keys.

There is no single method to get the keypair to your minion. If you are spooling up minions on EC2, you could pass them in using user_data or a cloud-init script. If you are handing them off to a team of developers for provisioning dev machines, you will need a secure file transfer.

Security Warning

Since the minion key is already accepted on the master, distributing the private key poses a potential security risk. A malicious party will have access to your entire state tree and other sensitive data.

Preseed the Minion with the keys

You will want to place the minion keys before starting the salt-minion daemon:

/etc/salt/pki/minion/minion.pem
/etc/salt/pki/minion/minion.pub

Once in place, you should be able to start salt-minion and run salt-call state.highstate or any other salt commands that require master authentication.

Salt Masterless Quickstart¶

Running a masterless salt-minion lets you use salt's configuration management for a single machine. It is also useful for testing out state trees before deploying to a production setup.

The only real difference in using a standalone minion is that instead of issuing commands with salt, we use the salt-call command, like this:

salt-call --local state.highstate

Bootstrap Salt Minion¶

First we need to install the salt minion. The salt-bootstrap script makes this incredibly easy for any OS with a Bourne shell. You can use it like this:

wget -O - http://bootstrap.saltstack.org | sudo sh

Or see the salt-bootstrap documentation for other one liners. Additionally, if you are using Vagrant to test out salt, the salty-vagrant tool will provision the VM for you.

Create State Tree¶

Now we build an example state tree. This is where the configuration is defined. For more in depth directions, see the tutorial.

Create the top.sls file:

/srv/salt/top.sls:

base:
  '*':
    - webserver

Create our webserver state tree:

/srv/salt/webserver.sls:

apache:               # ID declaration
  pkg:                # state declaration
    - installed       # function declaration

The only thing left is to provision our minion using the highstate command. Salt-call also gives us an easy way to give us verbose output:

salt-call --local state.highstate -l debug

The --local flag tells the salt-minion to look for the state tree in the local file system. Normally the minion copies the state tree from the master and executes it from there.

That's it, good luck!

Standalone Minion¶

Since the Salt minion contains such extensive functionality it can be useful to run it standalone. A standalone minion can be used to do a number of things:

Stand up a master server via States (Salting a Salt Master)
Use salt-call commands on a system without connectivity to a master
Masterless States, run states entirely from files local to the minion

Telling Salt Call to Run Masterless¶

The salt-call command is used to run module functions locally on a minion instead of executing them from the master. Normally the salt-call command checks into the master to retrieve file server and pillar data, but when running standalone salt-call needs to be instructed to not check the master for this data. To instruct the minion to not look for a master when running salt-call the file_client configuration option needs to be set. By default the file_client is set to remote so that the minion knows that file server and pillar data are to be gathered from the master. When setting the file_client option to local the minion is configured to not gather this data from the master.

file_client: local

Now the salt-call command will not look for a master and will assume that the local system has all of the file ad pillar resources.

Running States Masterless¶

The state system can be easily run without a Salt master, with all needed files local to the minion. To do this the minion configuration file needs to be set up to know how to return file_roots information like the master. The file_roots setting defaults to /srv/salt for the base environment just like on the master:

file_roots:
  base:
    - /srv/salt

Now set up the Salt State Tree, top file, and SLS modules in the same way that they would be set up on a master. Now, with the file_client option set to local and an available state tree then calls to functions in the state module will use the information in the file_roots on the minion instead of checking in with the master.

Remember that when creating a state tree on a minion there are no syntax or path changes needed, SLS modules written to be used from a master do not need to be modified in any way to work with a minion.

This makes it easy to "script" deployments with Salt states without having to set up a master, and allows for these SLS modules to be easily moved into a Salt master as the deployment grows.

Now the declared state can now be executed with:

salt-call state.highstate

Or the salt-call command can be executed with the --local flag, this makes it unnecessary to change the configuration file:

salt-call state.highstate --local

How Do I Use Salt States?¶

Simplicity, Simplicity, Simplicity

Many of the most powerful and useful engineering solutions are founded on simple principles. The Salt SLS system strives to do just that. K.I.S.S. (Keep It Stupidly Simple)

The core of the Salt State system is the SLS, or SaLt State file. The SLS is a representation of the state in which a system should be in, and is set up to contain this data in a simple format. This is often called configuration management.

Note

This is just the beginning of using states, make sure to read up on pillar Pillar next.

It is All Just Data¶

Before delving into the particulars, it will help to understand that the SLS file is just a data structure under the hood. While understanding that the SLS is just a data structure isn't critical for understanding and making use of Salt States, it should help bolster knowledge of where the real power is.

SLS files are therefore, in reality, just dictionaries, lists, strings, and numbers. By using this approach Salt can be much more flexible. As one writes more state files, it becomes clearer exactly what is being written. The result is a system that is easy to understand, yet grows with the needs of the admin or developer.

In the section titled "State Data Structures" a reference exists, explaining in depth how the data is laid out.

Default Data - YAML¶

By default Salt represents the SLS data in what is one of the simplest serialization formats available - YAML.

A typical SLS file will often look like this in YAML:

Note

These demos use some generic service and package names, different distributions often use different names for packages and services. For instance apache should be replaced with httpd on a Red Hat system. Salt uses the name of the init script, systemd name, upstart name etc. based on what the underlying service management for the platform. To get a list of the available service names on a platform execute the service.get_all salt function.

Information on how to make states work with multiple distributions is later in the tutorial.

apache:
  pkg:
    - installed
  service:
    - running
    - require:
      - pkg: apache

This SLS data will ensure that the package named apache is installed, and that the apache service is running. The components can be explained in a simple way.

The first line is the ID for a set of data, and it is called the ID Declaration. This ID sets the name of the thing that needs to be manipulated.

The second and fourth lines are the start of the State Declarations, so they are using the pkg and service states respectively. The pkg state manages a software package to be installed via the system's native package manager, and the service state manages a system daemon.

The third and fifth lines are the function to run. This function defines what state the named package and service should be in. Here, the package is to be installed, and the service should be running.

Finally, on line six, is the word require. This is called a Requisite Statement, and it makes sure that the Apache service is only started after a successful installation of the apache package.

Adding Configs and Users¶

When setting up a service like an Apache web server, many more components may need to be added. The Apache configuration file will most likely be managed, and a user and group may need to be set up.

apache:
  pkg:
    - installed
  service:
    - running
    - watch:
      - pkg: apache
      - file: /etc/httpd/conf/httpd.conf
      - user: apache
  user.present:
    - uid: 87
    - gid: 87
    - home: /var/www/html
    - shell: /bin/nologin
    - require:
      - group: apache
  group.present:
    - gid: 87
    - require:
      - pkg: apache

/etc/httpd/conf/httpd.conf:
  file.managed:
    - source: salt://apache/httpd.conf
    - user: root
    - group: root
    - mode: 644

This SLS data greatly extends the first example, and includes a config file, a user, a group and new requisite statement: watch.

Adding more states is easy, since the new user and group states are under the Apache ID, the user and group will be the Apache user and group. The require statements will make sure that the user will only be made after the group, and that the group will be made only after the Apache package is installed.

Next,the require statement under service was changed to watch, and is now watching 3 states instead of just one. The watch statement does the same thing as require, making sure that the other states run before running the state with a watch, but it adds an extra component. The watch statement will run the state's watcher function for any changes to the watched states. So if the package was updated, the config file changed, or the user uid modified, then the service state's watcher will be run. The service state's watcher just restarts the service, so in this case, a change in the config file will also trigger a restart of the respective service.

Moving Beyond a Single SLS¶

When setting up Salt States in a scalable manner, more than one SLS will need to be used. The above examples were in a single SLS file, but two or more SLS files can be combined to build out a State Tree. The above example also references a file with a strange source - salt://apache/httpd.conf. That file will need to be available as well.

The SLS files are laid out in a directory structure on the Salt master; an SLS is just a file and files to download are just files.

The Apache example would be laid out in the root of the Salt file server like this:

apache/init.sls
apache/httpd.conf

So the httpd.conf is just a file in the apache directory, and is referenced directly.

But when using more than one single SLS file, more components can be added to the toolkit. Consider this SSH example:

ssh/init.sls:

openssh-client:
  pkg.installed

/etc/ssh/ssh_config:
  file.managed:
    - user: root
    - group: root
    - mode: 644
    - source: salt://ssh/ssh_config
    - require:
      - pkg: openssh-client

ssh/server.sls:

include:
  - ssh

openssh-server:
  pkg.installed

sshd:
  service.running:
    - require:
      - pkg: openssh-client
      - pkg: openssh-server
      - file: /etc/ssh/banner
      - file: /etc/ssh/sshd_config

/etc/ssh/sshd_config:
  file.managed:
    - user: root
    - group: root
    - mode: 644
    - source: salt://ssh/sshd_config
    - require:
      - pkg: openssh-server

/etc/ssh/banner:
  file:
    - managed
    - user: root
    - group: root
    - mode: 644
    - source: salt://ssh/banner
    - require:
      - pkg: openssh-server

Note

Notice that we use two similar ways of denoting that a file is managed by Salt. In the /etc/ssh/sshd_config state section above, we use the file.managed state declaration whereas with the /etc/ssh/banner state section, we use the file state declaration and add a managed attribute to that state declaration. Both ways produce an identical result; the first way -- using file.managed -- is merely a shortcut.

Now our State Tree looks like this:

apache/init.sls
apache/httpd.conf
ssh/init.sls
ssh/server.sls
ssh/banner
ssh/ssh_config
ssh/sshd_config

This example now introduces the include statement. The include statement includes another SLS file so that components found in it can be required, watched or as will soon be demonstrated - extended.

The include statement allows for states to be cross linked. When an SLS has an include statement it is literally extended to include the contents of the included SLS files.

Note that some of the SLS files are called init.sls, while others are not. More info on what this means can be found in the States Tutorial.

Extending Included SLS Data¶

Sometimes SLS data needs to be extended. Perhaps the apache service needs to watch additional resources, or under certain circumstances a different file needs to be placed.

In these examples, the first will add a custom banner to ssh and the second will add more watchers to apache to include mod_python.

ssh/custom-server.sls:

include:
  - ssh.server

extend:
  /etc/ssh/banner:
    file:
      - source: salt://ssh/custom-banner

python/mod_python.sls:

include:
  - apache

extend:
  apache:
    service:
      - watch:
        - pkg: mod_python

mod_python:
  pkg.installed

The custom-server.sls file uses the extend statement to overwrite where the banner is being downloaded from, and therefore changing what file is being used to configure the banner.

In the new mod_python SLS the mod_python package is added, but more importantly the apache service was extended to also watch the mod_python package.

Using extend with require or watch

The extend statement works differently for require or watch. It appends to, rather than replacing the requisite component.

Understanding the Render System¶

Since SLS data is simply that (data), it does not need to be represented with YAML. Salt defaults to YAML because it is very straightforward and easy to learn and use. But the SLS files can be rendered from almost any imaginable medium, so long as a renderer module is provided.

The default rendering system is the yaml_jinja renderer. The yaml_jinja renderer will first pass the template through the Jinja2 templating system, and then through the YAML parser. The benefit here is that full programming constructs are available when creating SLS files.

Other renderers available are yaml_mako and yaml_wempy which each use the Mako or Wempy templating system respectively rather than the jinja templating system, and more notably, the pure Python or py and pydsl renderers. The py renderer allows for SLS files to be written in pure Python, allowing for the utmost level of flexibility and power when preparing SLS data; while the pydsl renderer provides a flexible, domain-specific language for authoring SLS data in Python.

Note

The templating engines described above aren't just available in SLS files. They can also be used in file.managed states, making file management much more dynamic and flexible. Some examples for using templates in managed files can be found in the documentation for the file states, as well as the MooseFS example below.

Getting to Know the Default - yaml_jinja¶

The default renderer - yaml_jinja, allows for use of the jinja templating system. A guide to the Jinja templating system can be found here: http://jinja.pocoo.org/docs

When working with renderers a few very useful bits of data are passed in. In the case of templating engine based renderers, three critical components are available, salt, grains, and pillar. The salt object allows for any Salt function to be called from within the template, and grains allows for the Grains to be accessed from within the template. A few examples:

apache/init.sls:

apache:
  pkg.installed:
    {% if grains['os'] == 'RedHat'%}
    - name: httpd
    {% endif %}
  service.running:
    {% if grains['os'] == 'RedHat'%}
    - name: httpd
    {% endif %}
    - watch:
      - pkg: apache
      - file: /etc/httpd/conf/httpd.conf
      - user: apache
  user.present:
    - uid: 87
    - gid: 87
    - home: /var/www/html
    - shell: /bin/nologin
    - require:
      - group: apache
  group.present:
    - gid: 87
    - require:
      - pkg: apache

/etc/httpd/conf/httpd.conf:
  file.managed:
    - source: salt://apache/httpd.conf
    - user: root
    - group: root
    - mode: 644

This example is simple. If the os grain states that the operating system is Red Hat, then the name of the Apache package and service needs to be httpd.

A more aggressive way to use Jinja can be found here, in a module to set up a MooseFS distributed filesystem chunkserver:

moosefs/chunk.sls:

include:
  - moosefs

{% for mnt in salt['cmd.run']('ls /dev/data/moose*').split() %}
/mnt/moose{{ mnt[-1] }}:
  mount.mounted:
    - device: {{ mnt }}
    - fstype: xfs
    - mkmnt: True
  file.directory:
    - user: mfs
    - group: mfs
    - require:
      - user: mfs
      - group: mfs
{% endfor %}

/etc/mfshdd.cfg:
  file.managed:
    - source: salt://moosefs/mfshdd.cfg
    - user: root
    - group: root
    - mode: 644
    - template: jinja
    - require:
      - pkg: mfs-chunkserver

/etc/mfschunkserver.cfg:
  file.managed:
    - source: salt://moosefs/mfschunkserver.cfg
    - user: root
    - group: root
    - mode: 644
    - template: jinja
    - require:
      - pkg: mfs-chunkserver

mfs-chunkserver:
  pkg:
    - installed
mfschunkserver:
  service:
    - running
    - require:
{% for mnt in salt['cmd.run']('ls /dev/data/moose*') %}
      - mount: /mnt/moose{{ mnt[-1] }}
      - file: /mnt/moose{{ mnt[-1] }}
{% endfor %}
      - file: /etc/mfschunkserver.cfg
      - file: /etc/mfshdd.cfg
      - file: /var/lib/mfs

This example shows much more of the available power of Jinja. Multiple for loops are used to dynamically detect available hard drives and set them up to be mounted, and the salt object is used multiple times to call shell commands to gather data.

Introducing the Python and the PyDSL Renderers¶

Sometimes the chosen default renderer might not have enough logical power to accomplish the needed task. When this happens, the Python renderer can be used. Normally a YAML renderer should be used for the majority of SLS files, but an SLS file set to use another renderer can be easily added to the tree.

This example shows a very basic Python SLS file:

python/django.sls:

#!py

def run():
    '''
    Install the django package
    '''
    return {'include': ['python'],
            'django': {'pkg': ['installed']}}

This is a very simple example; the first line has an SLS shebang that tells Salt to not use the default renderer, but to use the py renderer. Then the run function is defined, the return value from the run function must be a Salt friendly data structure, or better known as a Salt HighState data structure.

Alternatively, using the pydsl renderer, the above example can be written more succinctly as:

python/django.sls:

#!pydsl

include('python', delayed=True)
state('django').pkg.installed()

This Python examples would look like this if they were written in YAML:

include:
  - python

django:
  pkg.installed

This example clearly illustrates that; one, using the YAML renderer by default is a wise decision and two, unbridled power can be obtained where needed by using a pure Python SLS.

Running and debugging salt states.¶

Once the rules in an SLS are ready, they should be tested to ensure they work properly. To invoke these rules, simply execute salt '*' state.highstate on the command line. If you get back only hostnames with a : after, but no return, chances are there is a problem with one or more of the sls files. On the minion, use the salt-call command: salt-call state.highstate -l debug to examine the output for errors. This should help troubleshoot the issue. The minions can also be started in the foreground in debug mode: salt-minion -l debug.

Next Reading¶

With an understanding of states, the next recommendation is to become familiar with Salt's pillar interface:

Pillar Walkthrough

States tutorial, part 1¶

The purpose of this tutorial is to demonstrate how quickly you can configure a system to be managed by Salt States. For detailed information about the state system please refer to the full states reference.

This tutorial will walk you through using Salt to configure a minion to run the Apache HTTP server and to ensure the server is running.

Before continuing make sure you have a working Salt installation by following the installation and the configuration instructions.

Stuck?

There are many ways to get help from the Salt community including our mailing list and our IRC channel #salt.

Setting up the Salt State Tree¶

States are stored in text files on the master and transferred to the minions on demand via the master's File Server. The collection of state files make up the State Tree.

To start using a central state system in Salt, the Salt File Server must first be set up. Edit the master config file (file_roots) and uncomment the following lines:

file_roots:
  base:
    - /srv/salt

Note

If you are deploying on FreeBSD via ports, the file_roots path defaults to /usr/local/etc/salt/states.

Restart the Salt master in order to pick up this change:

pkill salt-master
salt-master -d

Preparing the Top File¶

On the master, in the directory uncommented in the previous step, (/srv/salt by default), create a new file called top.sls and add the following:

base:
  '*':
    - webserver

The top file is separated into environments (discussed later). The default environment is base. Under the base environment a collection of minion matches is defined; for now simply specify all hosts (*).

Targeting minions

The expressions can use any of the targeting mechanisms used by Salt — minions can be matched by glob, PCRE regular expression, or by grains. For example:

base:
  'os:Fedora':
    - match: grain
    - webserver

Create an `sls` module¶

In the same directory as the top file, create an empty file named webserver.sls, containing the following:

apache:                 # ID declaration
  pkg:                  # state declaration
    - installed         # function declaration

The first line, called the ID declaration, is an arbitrary identifier. In this case it defines the name of the package to be installed. NOTE: the package name for the Apache httpd web server may differ depending on OS or distro — for example, on Fedora it is httpd but on Debian/Ubuntu it is apache2.

The second line, called the state declaration, defines which of the Salt States we are using. In this example, we are using the pkg state to ensure that a given package is installed.

The third line, called the function declaration, defines which function in the pkg state module to call.

Renderers

States sls files can be written in many formats. Salt requires only a simple data structure and is not concerned with how that data structure is built. Templating languages and DSLs are a dime-a-dozen and everyone has a favorite.

Building the expected data structure is the job of Salt renderers and they are dead-simple to write.

In this tutorial we will be using YAML in Jinja2 templates, which is the default format. The default can be changed by editing renderer in the master configuration file.

Install the package¶

Next, let's run the state we created. Open a terminal on the master and run:

% salt '*' state.highstate

Our master is instructing all targeted minions to run state.highstate. When a minion executes a highstate call it will download the top file and attempt to match the expressions. When it does match an expression the modules listed for it will be downloaded, compiled, and executed.

Once completed, the minion will report back with a summary of all actions taken and all changes made.

SLS File Namespace

Note that in the example above, the SLS file webserver.sls was referred to simply as webserver. The namespace for SLS files follows a few simple rules:

The .sls is discarded (i.e. webserver.sls becomes webserver).
Subdirectories can be used for better organization.
1. Each subdirectory is represented by a dot.
2. webserver/dev.sls is referred to as webserver.dev.
A file called init.sls in a subdirectory is referred to by the path of the directory. So, webserver/init.sls is referred to as webserver.
If both webserver.sls and webserver/init.sls happen to exist, webserver/init.sls will be ignored and webserver.sls will be the file referred to as webserver.

Troubleshooting Salt

If the expected output isn't seen, the following tips can help to narrow down the problem.

Turn up logging

Salt can be quite chatty when you change the logging setting to debug:

salt-minion -l debug

Run the minion in the foreground

By not starting the minion in daemon mode (-d) one can view any output from the minion as it works:

salt-minion &

Increase the default timeout value when running salt. For example, to change the default timeout to 60 seconds:

salt -t 60

For best results, combine all three:

salt-minion -l debug &          # On the minion
salt '*' state.highstate -t 60  # On the master

Next steps¶

This tutorial focused on getting a simple Salt States configuration working. Part 2 will build on this example to cover more advanced sls syntax and will explore more of the states that ship with Salt.

States tutorial, part 2¶

Note

This tutorial builds on topics covered in part 1. It is recommended that you begin there.

In the last part of the Salt States tutorial we covered the basics of installing a package. We will now modify our webserver.sls file to have requirements, and use even more Salt States.

Call multiple States¶

You can specify multiple state declarations under an ID declaration. For example, a quick modification to our webserver.sls to also start Apache if it is not running:

apache:
  pkg:
    - installed
  service:
    - running
    - require:
      - pkg: apache

Try stopping Apache before running state.highstate once again and observe the output.

Expand the SLS module¶

As you have seen, SLS modules are appended with the file extension .sls and are referenced by name starting at the root of the state tree. An SLS module can be also defined as a directory. Demonstrate that now by creating a directory named webserver and moving and renaming webserver.sls to webserver/init.sls. Your state directory should now look like this:

|- top.sls
`- webserver/
   `- init.sls

Organizing SLS modules

You can place additional .sls files in a state file directory. This affords much cleaner organization of your state tree on the filesystem. For example, if we created a webserver/django.sls file that module would be referenced as webserver.django.

In addition, States provide powerful includes and extending functionality which we will cover in Part 3.

Require other states¶

We now have a working installation of Apache so let's add an HTML file to customize our website. It isn't exactly useful to have a website without a webserver so we don't want Salt to install our HTML file until Apache is installed and running. Include the following at the bottom of your webserver/init.sls file:

apache:
  pkg:
    - installed
  service:
    - running
    - require:
      - pkg: apache

/var/www/index.html:                        # ID declaration
  file:                                     # state declaration
    - managed                               # function
    - source: salt://webserver/index.html   # function arg
    - require:                              # requisite declaration
      - pkg: apache                         # requisite reference

line 9 is the ID declaration. In this example it is the location we want to install our custom HTML file. (Note: the default location that Apache serves may differ from the above on your OS or distro. /srv/www could also be a likely place to look.)

Line 10 the state declaration. This example uses the Salt file state.

Line 11 is the function declaration. The managed function will download a file from the master and install it in the location specified.

Line 12 is a function arg declaration which, in this example, passes the source argument to the managed function.

Line 13 is a requisite declaration.

Line 14 is a requisite reference which refers to a state and an ID. In this example, it is referring to the ID declaration from our example in part 1. This declaration tells Salt not to install the HTML file until Apache is installed.

Next, create the index.html file and save it in the webserver directory:

<html>
    <head><title>Salt rocks</title></head>
    <body>
        <h1>This file brought to you by Salt</h1>
    </body>
</html>

Last, call state.highstate again and the minion will fetch and execute the highstate as well as our HTML file from the master using Salt's File Server:

salt '*' state.highstate

Verify that Apache is now serving your custom HTML.

require vs. watch

There are two requisite declarations, “require” and “watch”. Not every state supports “watch”. The service state does support “watch” and will restart a service based on the watch condition.

For example, if you use Salt to install an Apache virtual host configuration file and want to restart Apache whenever that file is changed you could modify our Apache example from earlier as follows:

/etc/httpd/extra/httpd-vhosts.conf:
  file:
    - managed
    - source: salt://webserver/httpd-vhosts.conf

apache:
  pkg:
    - installed
  service:
    - running
    - watch:
      - file: /etc/httpd/extra/httpd-vhosts.conf
    - require:
      - pkg: apache

If the pkg and service names differ on your OS or distro of choice you can specify each one separately using a name declaration which explained in Part 3.

Next steps¶

In part 3 we will discuss how to use includes, extends and templating to make a more complete State Tree configuration.

States tutorial, part 3¶

Note

This tutorial builds on topics covered in part 1 and part 2. It is recommended that you begin there.

This part of the tutorial will cover more advanced templating and configuration techniques for sls files.

Templating SLS modules¶

SLS modules may require programming logic or inline execution. This is accomplished with module templating. The default module templating system used is Jinja2 and may be configured by changing the renderer value in the master config.

All states are passed through a templating system when they are initially read. To make use of the templating system, simply add some templating markup. An example of an sls module with templating markup may look like this:

{% for usr in 'moe','larry','curly' %}
{{ usr }}:
  user.present
{% endfor %}

This templated sls file once generated will look like this:

moe:
  user.present
larry:
  user.present
curly:
  user.present

Here's a more complex example:

{% for usr in 'moe','larry','curly' %}
{{ usr }}:
  group:
    - present
  user:
    - present
    - gid_from_name: True
    - require:
      - group: {{ usr }}
{% endfor %}

Using Grains in SLS modules¶

Often times a state will need to behave differently on different systems. Salt grains objects are made available in the template context. The grains can be used from within sls modules:

apache:
  pkg.installed:
    {% if grains['os'] == 'RedHat' %}
    - name: httpd
    {% elif grains['os'] == 'Ubuntu' %}
    - name: apache2
    {% endif %}

Calling Salt modules from templates¶

All of the Salt modules loaded by the minion are available within the templating system. This allows data to be gathered in real time on the target system. It also allows for shell commands to be run easily from within the sls modules.

The Salt module functions are also made available in the template context as salt:

moe:
  user:
    - present
    - gid: {{ salt['file.group_to_gid']('some_group_that_exists') }}

Note that for the above example to work, some_group_that_exists must exist before the state file is processed by the templating engine.

Below is an example that uses the network.hw_addr function to retrieve the MAC address for eth0:

salt['network.hw_addr']('eth0')

Advanced SLS module syntax¶

Lastly, we will cover some incredibly useful techniques for more complex State trees.

Include declaration ¶

A previous example showed how to spread a Salt tree across several files. Similarly, requisites span multiple files by using an include declaration. For example:

python/python-libs.sls:

python-dateutil:
  pkg.installed

python/django.sls:

include:
  - python.python-libs

django:
  pkg.installed:
    - require:
      - pkg: python-dateutil

Extend declaration ¶

You can modify previous declarations by using an extend declaration. For example the following modifies the Apache tree to also restart Apache when the vhosts file is changed:

apache/apache.sls:

apache:
  pkg.installed

apache/mywebsite.sls:

include:
  - apache.apache

extend:
  apache:
    service:
      - running
      - watch:
        - file: /etc/httpd/extra/httpd-vhosts.conf

/etc/httpd/extra/httpd-vhosts.conf:
  file.managed:
    - source: salt://apache/httpd-vhosts.conf

Using extend with require or watch

The extend statement works differently for require or watch. It appends to, rather than replacing the requisite component.

Name declaration ¶

You can override the ID declaration by using a name declaration. For example, the previous example is a bit more maintainable if rewritten as follows:

apache/mywebsite.sls:

include:
  - apache.apache

extend:
  apache:
    service:
      - running
      - watch:
        - file: mywebsite

mywebsite:
  file.managed:
    - name: /etc/httpd/extra/httpd-vhosts.conf
    - source: salt://apache/httpd-vhosts.conf

Names declaration ¶

Even more powerful is using a names declaration to override the ID declaration for multiple states at once. This often can remove the need for looping in a template. For example, the first example in this tutorial can be rewritten without the loop:

stooges:
  user.present:
    - names:
      - moe
      - larry
      - curly

Next steps¶

In part 4 we will discuss how to use salt's file_roots to set up a workflow in which states can be "promoted" from dev, to QA, to production.

States tutorial, part 4¶

Note

This tutorial builds on topics covered in part 1, part 2 and part 3. It is recommended that you begin there.

This part of the tutorial will show how to use salt's file_roots to set up a workflow in which states can be "promoted" from dev, to QA, to production.

Salt fileserver path inheritance¶

Salt's fileserver allows for more than one root directory per environment, like in the below example, which uses both a local directory and a secondary location shared to the salt master via NFS:

# In the master config file (/etc/salt/master)
file_roots:
  base:
    - /srv/salt
    - /mnt/salt-nfs/base

Salt's fileserver collapses the list of root directories into a single virtual environment containing all files from each root. If the same file exists at the same relative path in more than one root, then the top-most match "wins". For example, if /srv/salt/foo.txt and /mnt/salt-nfs/base/foo.txt both exist, then salt://foo.txt will point to /srv/salt/foo.txt.

Environment configuration¶

Configure a multiple-environment setup like so:

file_roots:
  base:
    - /srv/salt/prod
  qa:
    - /srv/salt/qa
    - /srv/salt/prod
  dev:
    - /srv/salt/dev
    - /srv/salt/qa
    - /srv/salt/prod

Given the path inheritance described above, files within /srv/salt/prod would be available in all environments. Files within /srv/salt/qa would be available in both qa, and dev. Finally, the files within /srv/salt/dev would only be available within the dev environment.

Based on the order in which the roots are defined, new files/states can be placed within /srv/salt/dev, and pushed out to the dev hosts for testing.

Those files/states can then be moved to the same relative path within /srv/salt/qa, and they are now available only in the dev and qa environments, allowing them to be pushed to QA hosts and tested.

Finally, if moved to the same relative path within /srv/salt/prod, the files are now available in all three environments.

Practical Example¶

As an example, consider a simple website, installed to /var/www/foobarcom. Below is a top.sls that can be used to deploy the website:

/srv/salt/prod/top.sls:

base:
  'web*prod*':
    - webserver.foobarcom
qa:
  'web*qa*':
    - webserver.foobarcom
dev:
  'web*dev*':
    - webserver.foobarcom

Using pillar, roles can be assigned to the hosts:

/srv/pillar/top.sls:

base:
  'web*prod*':
    - webserver.prod
  'web*qa*':
    - webserver.qa
  'web*dev*':
    - webserver.dev

/srv/pillar/webserver/prod.sls:

webserver_role: prod

/srv/pillar/webserver/qa.sls:

webserver_role: qa

/srv/pillar/webserver/dev.sls:

webserver_role: dev

And finally, the SLS to deploy the website:

/srv/salt/prod/webserver/foobarcom.sls:

{% if pillar.get('webserver_role', '') %}
/var/www/foobarcom:
  file.recurse:
    - source: salt://webserver/src/foobarcom
    - env: {{ pillar['webserver_role'] }}
    - user: www
    - group: www
    - dir_mode: 755
    - file_mode: 644
{% endif %}

Given the above SLS, the source for the website should initially be placed in /srv/salt/dev/webserver/src/foobarcom.

First, let's deploy to dev. Given the configuration in the top file, this can be done using state.highstate:

salt --pillar 'webserver_role:dev' state.highstate

However, in the event that it is not desirable to apply all states configured in the top file (which could be likely in more complex setups), it is possible to apply just the states for the foobarcom website, using state.sls:

salt --pillar 'webserver_role:dev' state.sls webserver.foobarcom

Once the site has been tested in dev, then the files can be moved from /srv/salt/dev/webserver/src/foobarcom to /srv/salt/qa/webserver/src/foobarcom, and deployed using the following:

salt --pillar 'webserver_role:qa' state.sls webserver.foobarcom

Finally, once the site has been tested in qa, then the files can be moved from /srv/salt/qa/webserver/src/foobarcom to /srv/salt/prod/webserver/src/foobarcom, and deployed using the following:

salt --pillar 'webserver_role:prod' state.sls webserver.foobarcom

Thanks to Salt's fileserver inheritance, even though the files have been moved to within /srv/salt/prod, they are still available from the same salt:// URI in both the qa and dev environments.

Continue learning¶

The best way to continue learning about Salt States is to read through the reference documentation and to look through examples of existing state trees. Many pre-configured state trees can be found on Github in the saltstack-formulas collection of repositories.

If you have any questions, suggestions, or just want to chat with other people who are using Salt, we have a very active community and we'd love to hear from you.

Salt Stack Walkthrough¶

Welcome!¶

Welcome to Salt Stack! I am excited that you are interested in Salt and starting down the path to better infrastructure management. I developed (and am continuing to develop) Salt with the goal of making the best software available to manage computers of almost any kind. I hope you enjoy working with Salt and that the software can solve your real world needs!

Thomas S Hatch
Salt creator and chief developer
CTO of Salt Stack, Inc.

Note

This is the first of a series of walkthroughs and serves as the best entry point for people new to Salt, after this be sure to read up on pillar and more on states:

Starting States

Pillar Walkthrough

Getting Started¶

What is Salt?¶

Salt is a different approach to infrastructure management, it is founded on the idea that high speed communication with large numbers of systems can open up new capabilities. This approach makes Salt a powerful multitasking system that can solve many specific problems in an infrastructure. The backbone of Salt is the remote execution engine, which creates a high speed, secure and bi-directional communication net for groups of systems. On top of this communication system Salt provides an extremely fast, flexible and easy to use configuration management system called Salt States.

This unique approach to management makes for a transparent control system that is not only amazingly easy to set up and use, but also capable of solving very complex problems in infrastructures; as will be explored in this walk through.

Salt is being used today by some of the largest infrastructures in the world and has a proven ability to scale to astounding proportions without modification. With the proven ability to scale out well beyond many tens of thousands of servers, Salt has also proven to be an excellent choice for small deployments as well, lowering compute and management overhead for infrastructures as small as just a few systems.

Installing Salt¶

Salt Stack has been made to be very easy to install and get started. Setting up Salt should be as easy as installing Salt via distribution packages on Linux or via the Windows installer. The installation documents cover specific platform installation in depth.

Starting Salt¶

Salt functions on a master/minion topology. A master server acts as a central control bus for the clients (called minions), and the minions connect back to the master.

Setting Up the Salt Master¶

Turning on the Salt Master is easy, just turn it on! The default configuration is suitable for the vast majority of installations. The Salt master can be controlled by the local Linux/Unix service manager:

On Systemd based platforms (OpenSuse, Fedora):

systemctl start salt-master

On Upstart based systems (Ubuntu, older Fedora/RHEL):

service salt-master start

On SysV Init systems (Debian, Gentoo etc.):

/etc/init.d/salt-master start

Or the master can be started directly on the command line:

salt-master -d

The Salt Master can also be started in the foreground in debug mode, thus greatly increasing the command output:

salt-master -l debug

The Salt Master needs to bind to 2 TCP network ports on the system, these ports are 4505 and 4506. For more in depth information on firewalling these ports, the firewall tutorial is available here.

Setting up a Salt Minion¶

Note

The Salt Minion can operate with or without a Salt Master. This walkthrough assumes that the minion will be connected to the master, for information on how to run a master-less minion please see the masterless quickstart guide:

Masterless Minion Quickstart

The Salt Minion only needs to be aware of one piece of information to run, the network location of the master. By default the minion will look for the DNS name salt for the master, making the easiest approach to set internal DNS to resolve the name salt back to the Salt Master IP. Otherwise the minion configuration file will need to be edited, edit the configuration option master to point to the DNS name or the IP of the Salt Master:

Note

The default location of the configuration files is /etc/salt. Most platforms adhere to this convention, but platforms such as FreeBSD and Microsoft Windows place this file in different locations.

/etc/salt/minion:

master: saltmaster.example.com

Now that the master can be found, start the minion in the same way as the master; with the platform init system, or via the command line directly:

As a daemon:

salt-minion -d

In the foreground in debug mode:

salt-minion -l debug

Now that the minion is started it will generate cryptographic keys and attempt to connect to the master. The next step is to venture back to the master server and accept the new minion's public key.

When the minion is started, it will generate an id value. This is the name by which the minion will attempt to authenticate to the master. The following steps are attempted, in order to try to find a value that is not localhost:

/etc/hostname is checked (non-Windows only) Note: Not used currently, will be as of version 0.17.0.
The Python function socket.getfqdn() is run
/etc/hosts (%WINDIR%\system32\drivers\etc\hosts on Windows hosts) is checked for hostnames that map to anything within 127.0.0.0/8.

If none of the above are able to produce an id which is not localhost, then a sorted list of IP addresses on the minion (excluding any within 127.0.0.0/8) is inspected. The first publicly-routable IP address is used, if there is one. Otherwise, the first privately-routable IP address is used.

If all else fails, then localhost is used as a fallback.

Note

Overriding the id

The minion id can be manually specified using the id parameter in the minion config file.

Using salt-key¶

Salt authenticates minions using public key encryption and authentication. For a minion to start accepting commands from the master the minion keys need to be accepted. The salt-key command is used to manage all of the keys on the master. To list the keys that are on the master run a salt-key list command:

salt-key -L

The keys that have been rejected, accepted and pending acceptance are listed. The easiest way to accept the minion key is to accept all pending keys:

salt-key -A

Note

Keys should be verified! The secure thing to do before accepting a key is to run salt-key -p minion-id to print the public key for the minion. This can then be compared against the minion's public key file, which is located (on the minion, of course) at /etc/salt/pki/minion/minion.pub.

On the master:

# salt-key -p foo.domain.com
Accepted Keys:
foo.domain.com:  -----BEGIN PUBLIC KEY-----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-----END PUBLIC KEY-----

On the minion:

# cat /etc/salt/pki/minion/minion.pub
-----BEGIN PUBLIC KEY-----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-----END PUBLIC KEY-----

Sending the First Commands¶

Now that the minion is connected to the master and authenticated, the master can start to command the minion. Salt commands allow for a vast set of functions to be executed and for specific minions and groups of minions to be targeted for execution. This makes the salt command very powerful, but the command is also very usable, and easy to understand.

The salt command is comprised of command options, target specification, the function to execute, and arguments to the function. A simple command to start with looks like this:

salt '*' test.ping

The * is the target, which specifies all minions, and test.ping tells the minion to run the test.ping function. The result of running this command will be the master instructing all of the minions to execute test.ping in parallel and return the result. This is not an actual ICMP ping, but rather a simple function which returns True. Using test.ping is a good way of confirming that a minion is connected.

Note

Each minion registers itself with a unique minion id. This id defaults to the minion's hostname, but can be explicitly defined in the minion config as well by using the id parameter.

Getting to Know the Functions¶

Salt comes with a vast library of functions available for execution, and Salt functions are self documenting. To see what functions are available on the minions execute the sys.doc function:

salt '*' sys.doc

This will display a very large list of available functions and documentation on them, this documentation is also available here.

These functions cover everything from shelling out to package management to manipulating database servers. They comprise a powerful system management API which is the backbone to Salt configuration management and many other aspects of Salt.

Note

Salt comes with many plugin systems. The functions that are available via the salt command are called Execution Modules.

Helpful Functions to Know¶

The cmd module contains functions to shell out on minions, such as cmd.run and cmd.run_all:

salt '*' cmd.run 'ls -l /etc'

The pkg functions automatically map local system package managers to the same salt functions. This means that pkg.install will install packages via yum on Red Hat based systems, apt on Debian systems, etc.:

salt '*' pkg.install vim

Note

Some custom Linux spins and derivatives of other distros are not properly detected by Salt. If the above command returns an error message saying that pkg.install is not available, then you may need to override the pkg provider. This process is explained here.

The network.interfaces function will list all interfaces on a minion, along with their IP addresses, netmasks, MAC addresses, etc:

salt '*' network.interfaces

`salt-call`¶

The examples so far have described running commands from the Master using the salt command, but when troubleshooting it can be more beneficial to login to the minion directly and use salt-call. Doing so allows you to see the minion log messages specific to the command you are running (which are not part of the return data you see when running the command from the Master using salt), making it unnecessary to tail the minion log. More information on salt-call and how to use it can be found here.

Grains¶

Salt uses a system called Grains to build up static data about minions. This data includes information about the operating system that is running, CPU architecture and much more. The grains system is used throughout Salt to deliver platform data to many components and to users.

Grains can also be statically set, this makes it easy to assign values to minions for grouping and managing. A common practice is to assign grains to minions to specify what the role or roles a minion might be. These static grains can be set in the minion configuration file or via the grains.setval function.

Targeting¶

Salt allows for minions to be targeted based on a wide range of criteria. The default targeting system uses globular expressions to match minions, hence if there are minions named larry1, larry2, curly1 and curly2, a glob of larry* will match larry1 and larry2, and a glob of *1 will match larry1 and curly1.

Many other targeting systems can be used other than globs, these systems include:

Regular Expressions: Target using PCRE compliant regular expressions
Grains: Target based on grains data: Targeting with Grains
Pillar: Target based on pillar data: Targeting with Pillar
IP: Target based on IP addr/subnet/range
Compound: Create logic to target based on multiple targets: Targeting with Compound
Nodegroup: Target with nodegroups: Targeting with Nodegroup

The concepts of targets are used on the command line with salt, but also function in many other areas as well, including the state system and the systems used for ACLs and user permissions.

Passing in Arguments¶

Many of the functions available accept arguments, these arguments can be passed in on the command line:

salt '*' pkg.install vim

This example passes the argument vim to the pkg.install function, since many functions can accept more complex input then just a string the arguments are parsed through YAML, allowing for more complex data to be sent on the command line:

salt '*' test.echo 'foo: bar'

In this case Salt translates the string 'foo: bar' into the dictionary "{'foo': 'bar'}"

Note

Any line that contains a newline will not be parsed by yaml.

Salt States¶

Now that the basics are covered the time has come to evaluate States. Salt States, or the State System is the component of Salt made for configuration management. The State system is a fully functional configuration management system which has been designed to be exceptionally powerful while still being simple to use, fast, lightweight, deterministic and with salty levels of flexibility.

The state system is already available with a basic salt setup, no additional configuration is required, states can be set up immediately.

Note

Before diving into the state system, a brief overview of how states are constructed will make many of the concepts clearer. Salt states are based on data modeling, and build on a low level data structure that is used to execute each state function. Then more logical layers are built on top of each other. The high layers of the state system which this tutorial will cover consists of everything that needs to be known to use states, the two high layers covered here are the sls layer and the highest layer highstate.

Again, knowing that there are many layers of data management, will help with understanding states, but they never need to be used. Just as understanding how a compiler functions when learning a programming language, understanding what is going on under the hood of a configuration management system will also prove to be a valuable asset.

The First SLS Formula¶

The state system is built on sls formulas, these formulas are built out in files on Salt's file server. To make a very basic sls formula open up a file under /srv/salt named vim.sls and get vim installed:

/srv/salt/vim.sls:

vim:
  pkg.installed

Now install vim on the minions by calling the sls directly:

salt '*' state.sls vim

This command will invoke the state system and run the named sls which was just created, vim.

Now, to beef up the vim sls formula, a vimrc can be added:

/srv/salt/vim.sls:

vim:
  pkg.installed

/etc/vimrc:
  file.managed:
    - source: salt://vimrc
    - mode: 644
    - user: root
    - group: root

Now the desired vimrc needs to be copied into the Salt file server to /srv/salt/vimrc, in Salt everything is a file, so no path redirection needs to be accounted for. The vimrc file is placed right next to the vim.sls file. The same command as above can be executed to all the vim sls formulas and now include managing the file.

Note

Salt does not need to be restarted/reloaded or have the master manipulated in any way when changing sls formulas, they are instantly available.

Adding Some Depth¶

Obviously maintaining sls formulas right in the root of the file server will not scale out to reasonably sized deployments. This is why more depth is required. Start by making an nginx formula a better way, make an nginx subdirectory and add an init.sls file:

/srv/salt/nginx/init.sls:

nginx:
  pkg:
    - installed
  service:
    - running
    - require:
      - pkg: nginx

A few things are introduced in this sls formula, first is the service statement which ensures that the nginx service is running, but the nginx service can't be started unless the package is installed, hence the require. The require statement makes sure that the required component is executed before and that it results in success.

Note

The require option belongs to a family of options called requisites. Requisites are a powerful component of Salt States, for more information on how requisites work and what is available see: Requisites Also evaluation ordering is available in Salt as well: Ordering States

Now this new sls formula has a special name, init.sls, when an sls formula is named init.sls it inherits the name of the directory path that contains it, so this formula can be referenced via the following command:

salt '*' state.sls nginx

Now that subdirectories can be used the vim.sls formula can be cleaned up, but to make things more flexible (and to illustrate another point of course), move the vim.sls and vimrc into a new subdirectory called edit and change the vim.sls file to reflect the change:

/srv/salt/edit/vim.sls:

vim:
  pkg.installed

/etc/vimrc:
  file.managed:
    - source: salt://edit/vimrc
    - mode: 644
    - user: root
    - group: root

The only change in the file is fixing the source path for the vimrc file. Now the formula is referenced as edit.vim because it resides in the edit subdirectory. Now the edit subdirectory can contain formulas for emacs, nano, joe or any other editor that may need to be deployed.

Next Reading¶

Two walkthroughs are specifically recommended at this point. First, a deeper run through States, followed by an explanation of Pillar.

An understanding of Pillar is extremely helpful in using States.

Getting Deeper Into States¶

Two more in-depth States tutorials exist, which delve much more deeply into States functionality.

Thomas' original states tutorial, How Do I Use Salt States?, covers much more to get off the ground with States.
The States Tutorial also provides a fantastic introduction.

These tutorials include much more in depth information including templating sls formulas etc.

So Much More!¶

This concludes the initial Salt walkthrough, but there are many more things to learn still! These documents will cover important core aspects of Salt:

A few more tutorials are also available:

This still is only scratching the surface, many components such as the reactor and event systems, extending Salt, modular components and more are not covered here. For an overview of all Salt features and documentation, look at the Table of Contents.

Access Control System¶

New in version 0.10.4.

Salt maintains a standard system used to open granular control to non administrative users to execute Salt commands. The access control system has been applied to all systems used to configure access to non administrative control interfaces in Salt.These interfaces include, the peer system, the external auth system and the client acl system.

The access control system mandated a standard configuration syntax used in all of the three aforementioned systems. While this adds functionality to the configuration in 0.10.4, it does not negate the old configuration.

Now specific functions can be opened up to specific minions from specific users in the case of external auth and client ACLs, and for specific minions in the case of the peer system.

The access controls are manifested using matchers in these configurations:

client_acl:
  fred:
    - web\*:
      - pkg.list_pkgs
      - test.*
      - apache.*

In the above example, fred is able to send commands only to minions which match the specified glob target. This can be expanded to include other functions for other minions based on standard targets.

external_auth:
  pam:
    dave:
      - test.ping
      - mongo\*:
        - network.*
      - log\*:
        - network.*
        - pkg.*
      - 'G@os:RedHat':
        - kmod.*
    steve:
      - .*

The above allows for all minions to be hit by test.ping by dave, and adds a few functions that dave can execute on other minions. It also allows steve unrestricted access to salt commands.

External Authentication System¶

Salt 0.10.4 comes with a fantastic new way to open up running Salt commands to users. This system allows for Salt itself to pass through authentication to any authentication system (The Unix PAM system was the first) to determine if a user has permission to execute a Salt command.

The external authentication system allows for specific users to be granted access to execute specific functions on specific minions. Access is configured in the master configuration file, and uses the new access control system:

external_auth:
  pam:
    thatch:
      - 'web*':
        - test.*
        - network.*
    steve:
      - .*

So, the above allows the user thatch to execute functions in the test and network modules on the minions that match the web* target. User steve is given unrestricted access to minion commands.

The external authentication system can then be used from the command line by any user on the same system as the master with the -a option:

$ salt -a pam web\* test.ping

The system will ask the user for the credentials required by the authentication system and then publish the command.

Tokens¶

With external authentication alone the authentication credentials will be required with every call to Salt. This can be alleviated with Salt tokens.

The tokens are short term authorizations and can be easily created by just adding a -T option when authenticating:

$ salt -T -a pam web\* test.ping

Now a token will be created that has a expiration of, by default, 12 hours. This token is stored in a file named .salt_token in the active user's home directory. Once the token is created, it is sent with all subsequent communications. The user authentication does not need to be entered again until the token expires. The token expiration time can be set in the Salt master config file.

Pillar of Salt¶

Pillar is an interface for Salt designed to offer global values that can be distributed to all minions. Pillar data is managed in a similar way as the Salt State Tree.

Pillar was added to Salt in version 0.9.8

Note

Storing sensitive data

Unlike state tree, pillar data is only available for the targeted minion specified by the matcher type. This makes it useful for storing sensitive data specific to a particular minion.

Declaring the Master Pillar¶

The Salt Master server maintains a pillar_roots setup that matches the structure of the file_roots used in the Salt file server. Like the Salt file server the pillar_roots option in the master config is based on environments mapping to directories. The pillar data is then mapped to minions based on matchers in a top file which is laid out in the same way as the state top file. Salt pillars can use the same matcher types as the standard top file.

The configuration for the pillar_roots in the master config file is identical in behavior and function as file_roots:

pillar_roots:
  base:
    - /srv/pillar

This example configuration declares that the base environment will be located in the /srv/pillar directory. The top file used matches the name of the top file used for States, and has the same structure:

/srv/pillar/top.sls

base:
  '*':
    - packages

This further example shows how to use other standard top matching types (grain matching is used in this example) to deliver specific salt pillar data to minions with different os grains:

dev:
  'os:Debian':
    - match: grain
    - servers

/srv/pillar/packages.sls

{% if grains['os'] == 'RedHat' %}
apache: httpd
git: git
{% elif grains['os'] == 'Debian' %}
apache: apache2
git: git-core
{% endif %}

Now this data can be used from within modules, renderers, State SLS files, and more via the shared pillar dict:

apache:
  pkg:
    - installed
    - name: {{ pillar['apache'] }}

git:
  pkg:
    - installed
    - name: {{ pillar['git'] }}

Note that you cannot just list key/value-information in top.sls.

Pillar namespace flattened¶

The separate pillar files all share the same namespace. Given a top.sls of:

base:
  '*':
    - packages
    - services

a packages.sls file of:

bind: bind9

and a services.sls file of:

bind: named

Then a request for the bind pillar will only return 'named'; the 'bind9' value is not available. It is better to structure your pillar files with more hierarchy. For example your package.sls file could look like:

packages:
  bind: bind9

Including Other Pillars¶

New in version 0.16.0.

Pillar SLS files may include other pillar files, similar to State files. Two syntaxes are available for this purpose. The simple form simply includes the additional pillar as if it were part of the same file:

include:
  - users

The full include form allows two additional options -- passing default values to the templating engine for the included pillar file as well as an optional key under which to nest the results of the included pillar:

include:
  - users:
      defaults:
        - sudo: ['bob', 'paul']
      key: users

With this form, the included file (users.sls) will be nested within the 'users' key of the compiled pillar. Additionally, the 'sudo' value will be available as a template variable to users.sls.

Viewing Minion Pillar¶

Once the pillar is set up the data can be viewed on the minion via the pillar module, the pillar module comes with two functions, pillar.items and and pillar.raw. pillar.items will return a freshly reloaded pillar and pillar.raw will return the current pillar without a refresh:

salt '*' pillar.items

Note

Prior to version 0.16.2, this function is named pillar.data. This function name is still supported for backwards compatibility.

Pillar "get" Function¶

New in version 0.14.0.

The pillar.get function works much in the same way as the get method in a python dict, but with an enhancement: nested dict components can be extracted using a : delimiter.

If a structure like this is in pillar:

foo:
  bar:
    baz: qux

Extracting it from the raw pillar in an sls formula or file template is done this way:

{{ pillar['foo']['bar']['baz'] }}

Now, with the new pillar.get function the data can be safely gathered and a default can be set, allowing the template to fall back if the value is not available:

{{ salt['pillar.get']('foo:bar:baz', 'qux') }}

This makes handling nested structures much easier.

Refreshing Pillar Data¶

When pillar data is changed on the master the minions need to refresh the data locally. This is done with the saltutil.refresh_pillar function.

salt '*' saltutil.refresh_pillar

This function triggers the minion to asynchronously refresh the pillar and will always return None.

Targeting with Pillar¶

Pillar data can be used when targeting minions. This allows for ultimate control and flexibility when targeting minions.

salt -I 'somekey:specialvalue' test.ping

Like with Grains, it is possible to use globbing as well as match nested values in Pillar, by adding colons for each level that is being traversed. The below example would match minions with a pillar named foo, which is a dict containing a key bar, with a value beginning with baz:

salt -I 'foo:bar:baz*' test.ping

Master Config In Pillar¶

For convenience the data stored in the master configuration file is made available in all minion's pillars. This makes global configuration of services and systems very easy but may not be desired if sensitive data is stored in the master configuration.

To disable the master config from being added to the pillar set pillar_opts to False:

pillar_opts: False

Master Tops System¶

In 0.10.4 the external_nodes system was upgraded to allow for modular subsystems to be used to generate the top file data for a highstate run on the master.

The old external_nodes option still works, but will be removed in the future in favor of the new master_tops option which uses the modular system instead. The master tops system contains a number of subsystems that are loaded via the Salt loader interfaces like modules, states, returners, runners, etc.

Using the new master_tops option is simple:

master_tops:
  ext_nodes: cobbler-external-nodes

for Cobbler or:

master_tops:
  reclass:
    inventory_base_uri: /etc/reclass
    classes_uri: roles

for Reclass.

Job Management¶

New in version 0.9.7.

Since Salt executes jobs running on many systems, Salt needs to be able to manage jobs running on many systems. As of Salt 0.9.7, the capability was added for more advanced job management.

The Minion proc System¶

The Salt Minions now maintain a proc directory in the Salt cachedir, the proc directory maintains files named after the executed job ID. These files contain the information about the current running jobs on the minion and allow for jobs to be looked up. This is located in the proc directory under the cachedir, with a default configuration it is under /var/cache/salt/proc.

Functions in the saltutil Module¶

Salt 0.9.7 introduced a few new functions to the saltutil module for managing jobs. These functions are:

running Returns the data of all running jobs that are found in the proc directory.
find_job Returns specific data about a certain job based on job id.
signal_job Allows for a given jid to be sent a signal.
term_job Sends a termination signal (SIGTERM, 15) to the process controlling the specified job.
kill_job Sends a kill signal (SIGKILL, 9) to the process controlling the specified job.

These functions make up the core of the back end used to manage jobs at the minion level.

The jobs Runner¶

A convenience runner front end and reporting system has been added as well. The jobs runner contains functions to make viewing data easier and cleaner.

The jobs runner contains a number of functions...

active¶

The active function runs saltutil.running on all minions and formats the return data about all running jobs in a much more usable and compact format. The active function will also compare jobs that have returned and jobs that are still running, making it easier to see what systems have completed a job and what systems are still being waited on.

# salt-run jobs.active

lookup_jid¶

When jobs are executed the return data is sent back to the master and cached. By default is is cached for 24 hours, but this can be configured via the keep_jobs option in the master configuration. Using the lookup_jid runner will display the same return data that the initial job invocation with the salt command would display.

# salt-run jobs.lookup_jid <job id number>

list_jobs¶

Before finding a historic job, it may be required to find the job id. list_jobs will parse the cached execution data and display all of the job data for jobs that have already, or partially returned.

# salt-run jobs.list_jobs

Salt Scheduling¶

In Salt versions greater than 0.12.0, the scheduling system allows incremental executions on minions or the master. The schedule system exposes the execution of any execution function on minions or any runner on the master.

To set up the scheduler on the master add the schedule option to the master config file.

To set up the scheduler on the minion add the schedule option to the minion config file or to the minion's pillar.

Note

The scheduler executes different functions on the master and minions. When running on the master the functions reference runner functions, when running on the minion the functions specify execution functions.

The schedule option defines jobs which execute at certain intervals. To set up a highstate to run on a minion every 60 minutes set this in the minion config or pillar:

schedule:
  highstate:
    function: state.highstate
    minutes: 60

Time intervals can be specified as seconds, minutes, hours, or days. Runner executions can also be specified on the master within the master configuration file:

schedule:
  overstate:
    function: state.over
    seconds: 35
    minutes: 30
    hours: 3

The above configuration will execute the state.over runner every 3 hours, 30 minutes and 35 seconds, or every 12,635 seconds.

Scheduler With Returner¶

The scheduler is also useful for tasks like gathering monitoring data about a minion, this schedule option will gather status data and send it to a mysql returner database:

schedule:
  uptime:
    function: status.uptime
    seconds: 60
    returner: mysql
  meminfo:
    function: status.meminfo
    minutes: 5
    returner: mysql

Since specifying the returner repeatedly can be tiresome, the schedule_returner option is available to specify one or a list of global returners to be used by the minions when scheduling.

Running the Salt Master as Unprivileged User¶

While the default setup runs the Salt Master as the root user, it is generally wise to run servers as an unprivileged user. In Salt 0.9.10 the management of the running user was greatly improved, the only change needed is to alter the option user in the master configuration file and all salt system components will be updated to function under the new user when the master is started.

If running a version older that 0.9.10 then a number of files need to be owned by the user intended to run the master:

# chown -R <user> /var/cache/salt
# chown -R <user> /var/log/salt
# chown -R <user> /etc/salt/pki

Troubleshooting¶

The intent of the troubleshooting section is to introduce solutions to a number of common issues encountered by users and the tools that are available to aid in developing States and Salt code.

Running in the Foreground¶

A great deal of information is available via the debug logging system, if you are having issues with minions connecting or not starting run the minion and/or master in the foreground:

salt-master -l debug
salt-minion -l debug

Anyone wanting to run Salt daemons via a process supervisor such as monit, runit, or supervisord, should omit the -d argument to the daemons and run them in the foreground.

What Ports do the Master and Minion Need Open?¶

No ports need to be opened up on each minion. For the master, TCP ports 4505 and 4506 need to be open. If you've put both your Salt master and minion in debug mode and don't see an acknowledgment that your minion has connected, it could very well be a firewall.

You can check port connectivity from the minion with the nc command:

nc -v -z salt.master.ip 4505
nc -v -z salt.master.ip 4506

There is also a firewall configuration document that might help as well.

If you've enabled the right TCP ports on your operating system or Linux distribution's firewall and still aren't seeing connections, check that no additional access control system such as SELinux or AppArmor is blocking Salt.

Using salt-call¶

The salt-call command was originally developed for aiding in the development of new Salt modules. Since then, many applications have been developed for running any Salt module locally on a minion. These range from the original intent of salt-call, development assistance, to gathering more verbose output from calls like state.highstate.

When creating your state tree, it is generally recommended to invoke state.highstate with salt-call. This displays far more information about the highstate execution than calling it remotely. For even more verbosity, increase the loglevel with the same argument as salt-minion:

salt-call -l debug state.highstate

The main difference between using salt and using salt-call is that salt-call is run from the minion, and it only runs the selected function on that minion. By contrast, salt is run from the master, and requires you to specify the minions on which to run the command using salt's targeting system.

Too many open files¶

The salt-master needs at least 2 sockets per host that connects to it, one for the Publisher and one for response port. Thus, large installations may, upon scaling up the number of minions accessing a given master, encounter:

12:45:29,289 [salt.master    ][INFO    ] Starting Salt worker process 38
Too many open files
sock != -1 (tcp_listener.cpp:335)

The solution to this would be to check the number of files allowed to be opened by the user running salt-master (root by default):

[root@salt-master ~]# ulimit -n
1024

And modify that value to be at least equal to the number of minions x 2. This setting can be changed in limits.conf as the nofile value(s), and activated upon new a login of the specified user.

So, an environment with 1800 minions, would need 1800 x 2 = 3600 as a minimum.

Salt Master Stops Responding¶

There are known bugs with ZeroMQ versions less than 2.1.11 which can cause the Salt master to not respond properly. If you're running a ZeroMQ version greater than or equal to 2.1.9, you can work around the bug by setting the sysctls net.core.rmem_max and net.core.wmem_max to 16777216. Next, set the third field in net.ipv4.tcp_rmem and net.ipv4.tcp_wmem to at least 16777216.

You can do it manually with something like:

# echo 16777216 > /proc/sys/net/core/rmem_max
# echo 16777216 > /proc/sys/net/core/wmem_max
# echo "4096 87380 16777216" > /proc/sys/net/ipv4/tcp_rmem
# echo "4096 87380 16777216" > /proc/sys/net/ipv4/tcp_wmem

Or with the following Salt state:

net.core.rmem_max:
  sysctl:
    - present
    - value: 16777216

net.core.wmem_max:
  sysctl:
    - present
    - value: 16777216

net.ipv4.tcp_rmem:
  sysctl:
    - present
    - value: 4096 87380 16777216

net.ipv4.tcp_wmem:
  sysctl:
    - present
    - value: 4096 87380 16777216

Salt and SELinux¶

Currently there are no SELinux policies for Salt. For the most part Salt runs without issue when SELinux is running in Enforcing mode. This is because when the minion executes as a daemon the type context is changed to initrc_t. The problem with SELinux arises when using salt-call or running the minion in the foreground, since the type context stays unconfined_t.

This problem is generally manifest in the rpm install scripts when using the pkg module. Until a full SELinux Policy is available for Salt the solution to this issue is to set the execution context of salt-call and salt-minion to rpm_exec_t:

# CentOS 5 and RHEL 5:
chcon -t system_u:system_r:rpm_exec_t:s0 /usr/bin/salt-minion
chcon -t system_u:system_r:rpm_exec_t:s0 /usr/bin/salt-call

# CentOS 6 and RHEL 6:
chcon system_u:object_r:rpm_exec_t:s0 /usr/bin/salt-minion
chcon system_u:object_r:rpm_exec_t:s0 /usr/bin/salt-call

This works well, because the rpm_exec_t context has very broad control over other types.

Red Hat Enterprise Linux 5¶

Salt requires Python 2.6 or 2.7. Red Hat Enterprise Linux 5 and its variants come with Python 2.4 installed by default. When installing on RHEL 5 from the EPEL repository this is handled for you. But, if you run Salt from git, be advised that its dependencies need to be installed from EPEL and that Salt needs to be run with the python26 executable.

Common YAML Gotchas¶

An extensive list of YAML idiosyncrasies has been compiled.

Live Python Debug Output¶

If the minion or master seems to be unresponsive, a SIGUSR1 can be passed to the processes to display where in the code they are running. If encountering a situation like this, this debug information can be invaluable. First make sure the master of minion are running in the foreground:

salt-master -l debug
salt-minion -l debug

The pass the signal to the master or minion when it seems to be unresponsive:

killall -SIGUSR1 salt-master
killall -SIGUSR1 salt-minion

When filing an issue or sending questions to the mailing list for a problem with an unresponsive daemon this information can be invaluable.

YAML Idiosyncrasies¶

One of Salt's strengths, the use of existing serialization systems for representing SLS data, can also backfire. YAML is a general purpose system and there are a number of things that would seem to make sense in an sls file that cause YAML issues. It is wise to be aware of these issues. While reports or running into them are generally rare they can still crop up at unexpected times.

Spaces vs Tabs¶

YAML uses spaces, period. Do not use tabs in your SLS files! If strange errors are coming up in rendering SLS files, make sure to check that no tabs have crept in! In Vim, after enabling search highlighting with: :set hlsearch, you can check with the following key sequence in normal mode(you can hit ESC twice to be sure): /, Ctrl-v, Tab, then hit Enter. Also, you can convert tabs to 2 spaces by these commands in Vim: :set tabstop=2 expandtab and then :retab.

Indentation¶

The suggested syntax for YAML files is to use 2 spaces for indentation, but YAML will follow whatever indentation system that the individual file uses. Indentation of two spaces works very well for SLS files given the fact that the data is uniform and not deeply nested.

Nested Dicts (key=value)¶

When dicts are more deeply nested, they no longer follow the same indentation logic. This is rarely something that comes up in Salt, since deeply nested options like these are discouraged when making State modules, but some do exist. A good example is the context and default options in the file.managed state:

/etc/http/conf/http.conf:
  file:
    - managed
    - source: salt://apache/http.conf
    - user: root
    - group: root
    - mode: 644
    - template: jinja
    - context:
        custom_var: "override"
    - defaults:
        custom_var: "default value"
        other_var: 123

Notice that the spacing used is 2 spaces, and that when defining the context and defaults options there is a 4 space indent. If only a 2 space indent is used then the information will not be loaded correctly. If using double spacing is not desirable, then a deeply nested dict can be declared with curly braces:

/etc/http/conf/http.conf:
  file:
    - managed
    - source: salt://apache/http.conf
    - user: root
    - group: root
    - mode: 644
    - template: jinja
    - context: {
      custom_var: "override" }
    - defaults: {
      custom_var: "default value",
      other_var: 123 }

True/False, Yes/No, On/Off¶

PyYAML will load these values as boolean True or False. Un-capitalized versions will also be loaded as booleans (true, false, yes, no, on, and off). This can be especially problematic when constructing Pillar data. Make sure that your Pillars which need to use the string versions of these values are enclosed in quotes.

Integers are Parsed as Integers¶

NOTE: This has been fixed in salt 0.10.0, as of this release passing an integer that is preceded by a 0 will be correctly parsed

When passing integers into an SLS file, they are passed as integers. This means that if a state accepts a string value and an integer is passed, that an integer will be sent. The solution here is to send the integer as a string.

This is best explained when setting the mode for a file:

/etc/vimrc:
  file:
    - managed
    - source: salt://edit/vimrc
    - user: root
    - group: root
    - mode: 644

Salt manages this well, since the mode is passed as 644, but if the mode is zero padded as 0644, then it is read by YAML as an integer and evaluated as an octal value, 0644 becomes 420. Therefore, if the file mode is preceded by a 0 then it needs to be passed as a string:

/etc/vimrc:
  file:
    - managed
    - source: salt://edit/vimrc
    - user: root
    - group: root
    - mode: '0644'

YAML does not like "Double Short Decs"¶

If I can find a way to make YAML accept "Double Short Decs" then I will, since I think that double short decs would be awesome. So what is a "Double Short Dec"? It is when you declare a multiple short decs in one ID. Here is a standard short dec, it works great:

vim:
  pkg.installed

The short dec means that there are no arguments to pass, so it is not required to add any arguments, and it can save space.

YAML though, gets upset when declaring multiple short decs, for the record...

THIS DOES NOT WORK:

vim:
  pkg.installed
  user.present

Similarly declaring a short dec in the same ID dec as a standard dec does not work either...

ALSO DOES NOT WORK:

fred:
  user.present
  ssh_auth.present:
    - name: AAAAB3NzaC...
    - user: fred
    - enc: ssh-dss
    - require:
      - user: fred

The correct way is to define them like this:

vim:
  pkg.installed: []
  user.present: []

fred:
  user.present: []
  ssh_auth.present:
    - name: AAAAB3NzaC...
    - user: fred
    - enc: ssh-dss
    - require:
      - user: fred

Alternatively, they can be defined the "old way", or with multiple "full decs":

vim:
  pkg:
    - installed
  user:
    - present

fred:
  user:
    - present
  ssh_auth:
    - present
    - name: AAAAB3NzaC...
    - user: fred
    - enc: ssh-dss
    - require:
      - user: fred

YAML support only plain ASCII¶

According to YAML specification, only ASCII characters can be used.

Within double-quotes, special characters may be represented with C-style escape sequences starting with a backslash ( \ ).

Examples:

- micro: "\u00b5"
- copyright: "\u00A9"
- A: "\x41"
- alpha: "\u0251"
- Alef: "\u05d0"

List of usable Unicode characters will help you to identify correct numbers.

Python can also be used to discover the Unicode number for a character:

repr(u"Text with wrong characters i need to figure out")

This shell command can find wrong characters in your SLS files:

find . -name '*.sls'  -exec  grep --color='auto' -P -n '[^\x00-\x7F]' \{} \;

Underscores stripped in Integer Definitions¶

If a definition only includes numbers and underscores, it is parsed by YAML as an integer and all underscores are stripped. To ensure the object becomes a string, it should be surrounded by quotes. More information here.

Here's an example:

>>> import yaml
>>> yaml.safe_load('2013_05_10')
20130510
>>> yaml.safe_load('"2013_05_10"')
'2013_05_10'

Community¶

Join the Salt!

There are many ways to participate in and communicate with the Salt community.

Salt has an active IRC channel and a mailing list.

Mailing List¶

Join the salt-users mailing list. It is the best place to ask questions about Salt and see whats going on with Salt development! The Salt mailing list is hosted by Google Groups. It is open to new members.

https://groups.google.com/forum/#!forum/salt-users

IRC¶

The #salt IRC channel is hosted on the popular Freenode network. You can use the Freenode webchat client right from your browser.

Logs of the IRC channel activity are being collected courtesy of Moritz Lenz.

Salt development¶

If you wish to discuss the development of Salt itself join us in #salt-devel.

Follow on Github¶

The Salt code is developed via Github. Follow Salt for constant updates on what is happening in Salt development:

https://github.com/saltstack/salt

The Red45 Blog¶

News and thoughts on Salt and related projects is often posted on Thomas' blog The Red45:

http://red45.wordpress.com/

Example Salt States¶

The official salt-states repository is: https://github.com/saltstack/salt-states

A few examples of salt states from the community:

Follow on ohloh¶

https://www.ohloh.net/p/salt

Developing Salt¶

There is a great need for contributions to salt and patches are welcome! The goal here is to make contributions clear, make sure there is a trail for where the code has come from, and most importantly, to give credit where credit is due!

There are a number of ways to contribute to salt development.

Sending a GitHub pull request¶

This is the preferred method for contributions. Simply create a GitHub fork, commit changes to the fork, and then open up a pull request.

The following is an example (from Open Comparison Contributing Docs ) of an efficient workflow for forking, cloning, branching, committing, and sending a pull request for a GitHub repository.

First, make a local clone of your GitHub fork of the salt GitHub repo and make edits and changes locally.

Then, create a new branch on your clone by entering the following commands:

git checkout -b fixed-broken-thing

Switched to a new branch 'fixed-broken-thing'

Choose a name for your branch that describes its purpose.

Now commit your changes to this new branch with the following command:

git commit -am 'description of my fixes for the broken thing'

Note

Using git commit -am, followed by a quoted string, both stages and commits all modified files in a single command. Depending on the nature of your changes, you may wish to stage and commit them separately. Also, note that if you wish to add newly-tracked files as part of your commit, they will not be caught using git commit -am and will need to be added using git add before committing.

Push your locally-committed changes back up to GitHub:

git push --set-upstream origin fixed-broken-thing

Now go look at your fork of the salt repo on the GitHub website. The new branch will now be listed under the "Source" tab where it says "Switch Branches". Select the new branch from this list, and then click the "Pull request" button.

Put in a descriptive comment, and include links to any project issues related to the pull request.

The repo managers will be notified of your pull request and it will be reviewed. If a reviewer asks for changes, just make the changes locally in the same local feature branch, push them to GitHub, then add a comment to the discussion section of the pull request.

Note

Travis-CI

To make reviewing pull requests easier for the maintainers, please enable Travis-CI on your fork. Salt is already configured, so simply follow the first 2 steps on the Travis-CI Getting Started Doc.

Keeping Salt Forks in Sync¶

Salt is advancing quickly. It is therefore critical to pull upstream changes from master into forks on a regular basis. Nothing is worse than putting in a days of hard work into a pull request only to have it rejected because it has diverged too far from master.

To pull in upstream changes:

# For ssh github
git remote add upstream git@github.com:saltstack/salt.git
git fetch upstream

# For https github
git remote add upstream https://github.com/saltstack/salt.git
git fetch upstream

To check the log to be sure that you actually want the changes, run the following before merging:

git log upstream/develop

Then to accept the changes and merge into the current branch:

git merge upstream/develop

For more info, see GitHub Fork a Repo Guide or Open Comparison Contributing Docs

Posting patches to the mailing list¶

Patches will also be accepted by email. Format patches using git format-patch and send them to the Salt users mailing list. The contributor will then get credit for the patch, and the Salt community will have an archive of the patch and a place for discussion.

Installing Salt for development¶

Clone the repository using:

git clone https://github.com/saltstack/salt

Note

Running a self-contained development version¶

During development it is easiest to be able to run the Salt master and minion that are installed in the virtualenv you created above, and also to have all the configuration, log, and cache files contained in the virtualenv as well.

Copy the master and minion config files into your virtualenv:

mkdir -p /path/to/your/virtualenv/etc/salt
cp ./salt/conf/master /path/to/your/virtualenv/etc/salt/master
cp ./salt/conf/minion /path/to/your/virtualenv/etc/salt/minion

Edit the master config file:

Uncomment and change the user: root value to your own user.
Uncomment and change the root_dir: / value to point to /path/to/your/virtualenv.
If you are running version 0.11.1 or older, uncomment and change the pidfile: /var/run/salt-master.pid value to point to /path/to/your/virtualenv/salt-master.pid.
If you are also running a non-development version of Salt you will have to change the publish_port and ret_port values as well.

Edit the minion config file:

Repeat the edits you made in the master config for the user and root_dir values as well as any port changes.
If you are running version 0.11.1 or older, uncomment and change the pidfile: /var/run/salt-minion.pid value to point to /path/to/your/virtualenv/salt-minion.pid.
Uncomment and change the master: salt value to point at localhost.
Uncomment and change the id: value to something descriptive like "saltdev". This isn't strictly necessary but it will serve as a reminder of which Salt installation you are working with.

Note

Using salt-call with a Standalone Minion

If you plan to run salt-call with this self-contained development environment in a masterless setup, you should invoke salt-call with -c /path/to/your/virtualenv/etc/salt so that salt can find the minion config file. Without the -c option, Salt finds its config files in /etc/salt.

Start the master and minion, accept the minion's key, and verify your local Salt installation is working:

cd /path/to/your/virtualenv
salt-master -c ./etc/salt -d
salt-minion -c ./etc/salt -d
salt-key -c ./etc/salt -L
salt-key -c ./etc/salt -A
salt -c ./etc/salt '*' test.ping

Running the master and minion in debug mode can be helpful when developing. To do this, add -l debug to the calls to salt-master and salt-minion. If you would like to log to the console instead of to the log file, remove the -d.

Once the minion starts, you may see an error like the following:

zmq.core.error.ZMQError: ipc path "/path/to/your/virtualenv/var/run/salt/minion/minion_event_7824dcbcfd7a8f6755939af70b96249f_pub.ipc" is longer than 107 characters (sizeof(sockaddr_un.sun_path)).

This means the the path to the socket the minion is using is too long. This is a system limitation, so the only workaround is to reduce the length of this path. This can be done in a couple different ways:

Create your virtualenv in a path that is short enough.
Edit the sock_dir minion config variable and reduce its length. Remember that this path is relative to the value you set in root_dir.

NOTE: The socket path is limited to 107 characters on Solaris and Linux, and 103 characters on BSD-based systems.

Note

File descriptor limits

Ensure that the system open file limit is raised to at least 2047:

# check your current limit
ulimit -n

# raise the limit. persists only until reboot
# use 'limit descriptors 2047' for c-shell
ulimit -n 2047

To set file descriptors on OSX, refer to the OS X Installation instructions.

Using easy_install to Install Salt¶

If you are installing using easy_install, you will need to define a USE_SETUPTOOLS environment variable, otherwise dependencies will not be installed:

USE_SETUPTOOLS=1 easy_install salt

Running the tests¶

You will need mock to run the tests:

pip install mock

If you are on Python < 2.7 then you will also need unittest2:

pip install unittest2

Finally you use setup.py to run the tests with the following command:

./setup.py test

For greater control while running the tests, please try:

./tests/runtests.py -h

Editing and previewing the documentation¶

You need sphinx-build command to build the docs. In Debian/Ubuntu this is provided in the python-sphinx package. Sphinx can also be installed to a virtualenv using pip:

pip install Sphinx

Change to salt documentation directory, then:

cd doc; make html

This will build the HTML docs. Run make without any arguments to see the available make targets, which include html, man, and text.
The docs then are built within the docs/_build/ folder. To update the docs after making changes, run make again.
The docs use reStructuredText for markup. See a live demo at http://rst.ninjs.org/.
The help information on each module or state is culled from the python code that runs for that piece. Find them in salt/modules/ or salt/states/.
To build the docs on Arch Linux, the python2-sphinx package is required. Additionally, it is necessary to tell make where to find the proper sphinx-build binary, like so:

make SPHINXBUILD=sphinx-build2 html

To build the docs on RHEL/CentOS 6, the python-sphinx10 package must be installed from EPEL, and the following make command must be used:

make SPHINXBUILD=sphinx-1.0-build html

Salt Based Projects¶

A number of unofficial open source projects, based on Salt, or written to enhance Salt have been created.

Salt Sandbox¶

Created by Aaron Bull Schaefer, aka "elasticdog".

https://github.com/elasticdog/salt-sandbox

Salt Sandbox is a multi-VM Vagrant-based Salt development environment used for creating and testing new Salt state modules outside of your production environment. It's also a great way to learn firsthand about Salt and its remote execution capabilities.

Salt Sandbox will set up three separate virtual machines:

salt.example.com - the Salt master server
minion1.example.com - the first Salt minion machine
minion2.example.com - the second Salt minion machine

These VMs can be used in conjunction to segregate and test your modules based on node groups, top file environments, grain values, etc. You can even test modules on different Linux distributions or release versions to better match your production infrastructure.

Salt Event System¶

Salt 0.9.10 introduced the Salt Event System. This system is used to fire off events enabling third party applications or external processes to react to behavior within Salt.

The event system is comprised of a few components, the event sockets which publish events, and the event library which can listen to events and send events into the salt system.

Listening for Events¶

The event system is accessed via the event library and can only be accessed by the same system user that Salt is running as. To listen to events a SaltEvent object needs to be created and then the get_event function needs to be run. The SaltEvent object needs to know the location that the Salt Unix sockets are kept. In the configuration this is the sock_dir option. The sock_dir option defaults to "/var/run/salt/master" on most systems.

The following code will check for a single event:

import salt.utils.event

event = salt.utils.event.MasterEvent('/var/run/salt/master')

data = event.get_event()

Events will also use a "tag". A "tag" allows for events to be filtered. By default all events will be returned, but if only authentication events are desired, then pass the tag "auth". Also, the get_event method has a default poll time assigned of 5 seconds, to change this time set the "wait" option. This example will only listen for auth events and will wait for 10 seconds instead of the default 5.

import salt.utils.event

event = salt.utils.event.MasterEvent('/var/run/salt/master')

data = event.get_event(wait=10, tag='auth')

Instead of looking for a single event, the iter_events method can be used to make a generator which will continually yield salt events. The iter_events method also accepts a tag, but not a wait time:

import salt.utils.event

event = salt.utils.event.MasterEvent('/var/run/salt/master')

for data in event.iter_events(tag='auth'):
    print(data)

Firing Events¶

It is possible to fire events on either the minion's local bus, or to fire events intended for the master. To fire a local event from the minion, on the command line:

salt-call event.fire 'message to be sent in the event' 'tag'

To fire an event to be sent to the master, from the minion:

salt-call event.fire_master 'message for the master' 'tag'

If a process is listening on the minion, it may be useful for a user on the master to fire an event to it:

salt minionname event.fire 'message for the minion' 'tag'

Firing Events From Code¶

Events can be very useful when writing execution modules, in order to inform various processes on the master when a certain task has taken place. In Salt versions previous to 0.17.0, the basic code looks like:

# Import the proper library
import salt.utils.event
# Fire deploy action
sock_dir = '/var/run/salt/minion'
event = salt.utils.event.SaltEvent('master', sock_dir)
event.fire_event('Message to be sent', 'tag')

In Salt version 0.17.0, the ability to send a payload with a more complex data structure than a string was added. When using this interface, a Python dictionary should be sent instead.

# Import the proper library
import salt.utils.event
# Fire deploy action
sock_dir = '/var/run/salt/minion'
payload = {'sample-msg': 'this is a test',
           'example': 'this is the same test'}
event = salt.utils.event.SaltEvent('master', sock_dir)
event.fire_event(payload, 'tag')

It should be noted that this code can be used in 3rd party applications as well. So long as the salt-minion process is running, the minion socket can be used:

sock_dir = '/var/run/salt/minion'

So long as the salt-master process is running, the master socket can be used:

sock_dir = '/var/run/salt/master'

This allows 3rd party applications to harness the power of the Salt event bus programmatically, without having to make other calls to Salt. A 3rd party process can listen to the event bus on the master, and another 3rd party process can fire events to the process on the master, which Salt will happily pass along.

The Salt Mine¶

Granted, it took a while for this name to be used in Salt, but version 0.15.0 introduces a new system to Salt called the Salt Mine.

The Salt Mine is used to bridge the gap between setting static variables and gathering live data. The Salt mine is used to collect arbitrary data from minions and store it on the master. This data is then made available to all minions via the mine module.

The data is gathered on the minion and sent back to the master where only the most recent data is maintained (if long term data is required use returners or the external job cache).

Mine Functions¶

To enable the Salt Mine the mine_functions option needs to be applied to a minion. This option can be applied via the minion's configuration file, or the minion's pillar. The mine_functions option dictates what functions are being executed and allows for arguments to be passed in:

mine_functions:
  network.interfaces: []
  test.ping: []

Mine Interval¶

The Salt Mine functions are executed when the minion starts and at a given interval by the scheduler. The default interval is every 60 minutes and can be adjusted for the minion via the mine_interval option:

mine_interval: 60

Salt Virt - The Salt Stack Cloud Controller¶

The Salt Virt cloud controller capability was initial added to Salt in version 0.14.0 as an alpha technology.

The initial Salt Virt system supports core cloud operations:

Virtual machine deployment
Inspection of deployed VMs
Virtual machine migration
Network profiling
Automatic VM integration with all aspects of Salt
Image Pre-seeding

Many features are currently under development to enhance the capabilities of the Salt Virt systems.

Note

It is noteworthy that Salt was originally developed with the intent of using the Salt communication system as the backbone to a cloud controller. This means that the Salt Virt system is not an afterthought, simply a system that took the back seat to other development. The original attempt to develop the cloud control aspects of Salt was a project called butter. This project never took off, but was functional and proves the early viability of Salt to be a cloud controller.

Salt Virt Tutorial¶

A tutorial about how to get Salt Virt up and running has been added to the tutorial section:

Cloud Controller Tutorial

The Salt Virt Runner¶

The point of interaction with the cloud controller is the virt runner. The virt runner comes with routines to execute specific virtual machine routines.

Reference documentation for the virt runner is available with the runner module documentation:

Virt Runner Reference

Based on Live State Data¶

The Salt Virt system is based on using Salt to query live data about hypervisors and then using the data gathered to make decisions about cloud operations. This means that no external resources are required to run Salt Virt, and that the information gathered about the cloud is live and accurate.

Virtual Machine Network Profiles¶

Salt Virt allows for the network devices created for deployed virtual machines to be finely configured. The configuration is a simple data structure which is read from the config.option function, meaning that the configuration can be stored in the minion config file, the master config file, or the minion's pillar.

This configuration option is called virt.nic. By default the virt.nic option is empty but defaults to a data structure which looks like this:

virt.nic:
  default:
    eth0:
      bridge: br0
      model: virtio

Note

The model does not need to be defined, Salt will default to the optimal model used by the underlying hypervisor, in the case of kvm this model is virtio

This configuration sets up a network profile called default. The default profile creates a single Ethernet device on the virtual machine that is bridged to the hypervisor's br0 interface. This default setup does not require setting up the virt.nic configuration, and is the reason why a default install only requires setting up the br0 bridge device on the hypervisor.

Define More Profiles¶

Many environments will require more complex network profiles and may require more than one profile, this can be easily accomplished:

virt.nic:
  dual:
    eth0:
      bridge: service_br
    eth1:
      bridge: storage_br
  single:
    eth0:
      bridge: service_br
  triple:
    eth0:
      bridge: service_br
    eth1:
      bridge: storage_br
    eth2:
      bridge: dmz_br
  all:
    eth0:
      bridge: service_br
    eth1:
      bridge: storage_br
    eth2:
      bridge: dmz_br
    eth3:
      bridge: database_br
  dmz:
    eth0:
      bridge: service_br
    eth1:
      bridge: dmz_br
  database:
    eth0:
      bridge: service_br
    eth1:
      bridge: database_br

This configuration allows for one of six profiles to be selected, allowing virtual machines to be created which attach to different network depending on the needs of the deployed vm.

Salt SSH¶

In version 0.17.0 of Salt a new transport system was introduced, the ability to use SSH for Salt communication. This addition allows for Salt routines to be executed on remote systems entirely through ssh, bypassing the need for a Salt Minion to be running on the remote systems and the need for a Salt Master.

Note

The Salt SSH system does not supercede the standard Salt communication systems, it simply offers an SSH based alternative that does not require ZeroMQ and a remote agent. Be aware that since all communication with Salt SSH is executed via SSH it is substantially slower than standard Salt with ZeroMQ.

Salt SSH is very easy to use, simply set up a basic roster file of the systems to connect to and run salt-ssh commands in a similar way as standard salt commands.

Salt SSH Roster¶

The roster system in Salt allows for remote minions to be easily defined.

Note

See the Roster documentation for more details.

Simply create the roster file, the default location is /etc/salt/roster:

web1: 192.168.42.1

This is a very basic roster file where a Salt ID is being assigned to an IP address. A more elaborate roster can be created:

web1:
  host: 192.168.42.1 # The IP addr or DNS hostname
  user: fred         # Remote executions will be executed as user fred
  passwd: foobarbaz  # The password to use for login, if omitted, keys are used
  sudo: True         # Whether to sudo to root, not enabled by default
web2:
  host: 192.168.42.2

Calling Salt SSH¶

The salt-ssh command can be easily executed in the same was as a salt command:

salt-ssh '*' test.ping

Commands with salt-ssh follow the same syntax as the salt command.

The standard salt functions are available! The output is the same as salt and many of the same flags are available.

Raw Shell Calls¶

By default salt-ssh runs Salt execution modules on the remote system, but salt-ssh can also execute raw shell commands:

salt-ssh '*' -r 'ifconfig'

States Via Salt SSH¶

The Salt State system can also be used with salt-ssh. The state system abstracts the same interface to the user in salt-ssh as it does when using standard salt. The intent is that Salt Formulas defined for standard salt will work seamlessly with salt-ssh as vis-versa.

The standard Salt States walkthroughs function by simply replacing salt commands with salt-ssh.

Targeting with Salt SSH¶

Due to the fact that the targeting approach differs in salt-ssh, only glob and regex targets are supported as of this writing, the remaining target systems still need to be implemented.

Salt Rosters¶

Salt rosters are plugable systems added in Salt 0.17.0 to facilitate the salt-ssh system. The roster system was created because salt-ssh needs a means to identify which systems need to be targeted for execution.

Note

The Roster System is not needed or used in standard Salt because the master does not need to be initially aware of target systems, since the Salt Minion checks itself into the master.

Since the roster system is pluggable, it can be easily augmented to attach to any existing systems to gather information about what servers are presently available and should be attached to by salt-ssh.

How Rosters Work¶

The roster system compiles a data structure internally refered to as targets. The targets is a list of target systems and attributes about how to connect to said systems. The only requirement for a roster module in Salt is to return the targets data structure.

Targets Data¶

The information which can be stored in a roster target is the following:

<Salt ID>:   # The id to reference the target system with
    host:    # The IP address or DNS name of the remote host
    user:    # The user to log in as
    passwd:  # The password to log in with

Running The Tests¶

To run the tests, use tests/runtests.py, see --help for more info.

Examples:

To run all tests: sudo ./tests/runtests.py
Run unit tests only: sudo ./tests/runtests.py --unit-tests

You will need 'mock' (https://pypi.python.org/pypi/mock) in addition to salt requirements in order to run the tests.

Writing Tests¶

Salt uses a test platform to verify functionality of components in a simple way. Two testing systems exist to enable testing salt functions in somewhat real environments. The two subsystems available are integration tests and unit tests.

Salt uses the python standard library unittest2 system for testing.

Integration Tests¶

The integration tests start up a number of salt daemons to test functionality in a live environment. These daemons include 2 salt masters, 1 syndic and 2 minions. This allows for the syndic interface to be tested and master/minion communication to be verified. All of the integration tests are executed as live salt commands sent through the started daemons.

Writing integration tests

Integration tests are particularly good at testing modules, states and shell commands.

Unit Tests¶

Direct unit tests are also available, these tests are good for internal functions.

Integration Tests¶

The Salt integration tests come with a number of classes and methods which allow for components to be easily tested. These classes are generally inherited from and provide specific methods for hooking into the running integration test environment created by the integration tests.

It is noteworthy that since integration tests validate against a running environment that they are generally the preferred means to write tests.

The integration system is all located under tests/integration in the Salt source tree.

Integration Classes¶

The integration classes are located in tests/integration/__init__.py and can be extended therein. There are three classes available to extend:

ModuleCase¶

Used to define executions run via the master to minions and to call single modules and states.

The available methods are as follows:

run_function:: Run a single salt function and condition the return down to match the behavior of the raw function call. This will run the command and only return the results from a single minion to verify.
state_result:: Return the result data from a single state return
run_state:: Run the state.single command and return the state return structure

SyndicCase¶

Used to execute remote commands via a syndic, only used to verify the capabilities of the Syndic.

The available methods are as follows:

run_function:: Run a single salt function and condition the return down to match the behavior of the raw function call. This will run the command and only return the results from a single minion to verify.

ShellCase¶

Shell out to the scripts which ship with Salt.

The available methods are as follows:

run_script:: Execute a salt script with the given argument string
run_salt:: Execute the salt command, pass in the argument string as it would be passed on the command line.
run_run:: Execute the salt-run command, pass in the argument string as it would be passed on the command line.
run_run_plus:: Execute Salt run and the salt run function and return the data from each in a dict
run_key:: Execute the salt-key command, pass in the argument string as it would be passed on the command line.
run_cp:: Execute salt-cp, pass in the argument string as it would be passed on the command line.
run_call:: Execute salt-call, pass in the argument string as it would be passed on the command line.

Examples¶

Module Example via ModuleCase Class¶

Import the integration module, this module is already added to the python path by the test execution. Inherit from the integration.ModuleCase class. The tests that execute against salt modules should be placed in the tests/integration/modules directory so that they will be detected by the test system.

Now the workhorse method run_function can be used to test a module:

import os
import integration


class TestModuleTest(integration.ModuleCase):
    '''
    Validate the test module
    '''
    def test_ping(self):
        '''
        test.ping
        '''
        self.assertTrue(self.run_function('test.ping'))

    def test_echo(self):
        '''
        test.echo
        '''
        self.assertEqual(self.run_function('test.echo', ['text']), 'text')

ModuleCase can also be used to test states, when testing states place the test module in the tests/integration/states directory. The state_result and the run_state methods are the workhorse here:

import os
import shutil
import integration

HFILE = os.path.join(integration.TMP, 'hosts')

class HostTest(integration.ModuleCase):
    '''
    Validate the host state
    '''

    def setUp(self):
        shutil.copyfile(os.path.join(integration.FILES, 'hosts'), HFILE)
        super(HostTest, self).setUp()

    def tearDown(self):
        if os.path.exists(HFILE):
            os.remove(HFILE)
        super(HostTest, self).tearDown()

    def test_present(self):
        '''
        host.present
        '''
        name = 'spam.bacon'
        ip = '10.10.10.10'
        ret = self.run_state('host.present', name=name, ip=ip)
        result = self.state_result(ret)
        self.assertTrue(result)
        with open(HFILE) as fp_:
            output = fp_.read()
            self.assertIn('{0}\t\t{1}'.format(ip, name), output)

The above example also demonstrates using the integration files and the integration state tree. The variable integration.FILES will point to the directory used to store files that can be used or added to to help enable tests that require files. The location integration.TMP can also be used to store temporary files that the test system will clean up when the execution finishes.

The integration state tree can be found at tests/integration/files/file/base. This is where the referenced host.present sls file resides.

Shell Example via ShellCase¶

Validating the shell commands can be done via shell tests. Here are some examples:

import sys
import shutil
import tempfile

import integration

class KeyTest(integration.ShellCase):
    '''
    Test salt-key script
    '''

    _call_binary_ = 'salt-key'

    def test_list(self):
        '''
        test salt-key -L
        '''
        data = self.run_key('-L')
        expect = [
                'Unaccepted Keys:',
                'Accepted Keys:',
                'minion',
                'sub_minion',
                'Rejected:', '']
        self.assertEqual(data, expect)

This example verifies that the salt-key command executes and returns as expected by making use of the run_key method.

All shell tests should be placed in the tests/integraion/shell directory.

Reactor System¶

Salt version 0.11.0 introduced the reactor system. The premise behind the reactor system is that with Salt's events and the ability to execute commands, a logic engine could be put in place to allow events to trigger actions, or more accurately, reactions.

This system binds sls files to event tags on the master. These sls files then define reactions. This means that the reactor system has two parts. First, the reactor option needs to be set in the master configuration file. The reactor option allows for event tags to be associated with sls reaction files. Second, these reaction files use highdata (like the state system) to define reactions to be executed.

Event System¶

A basic understanding of the event system is required to understand reactors. The event system is a local ZeroMQ PUB interface which fires salt events. This event bus is an open system used for sending information notifying Salt and other systems about operations.

The event system fires events with a very specific criteria. Every event has a tag which is comprised of a maximum of 20 characters. Event tags allow for fast top level filtering of events. In addition to the tag, each event has a data structure. This data structure is a dict, which contains information about the event.

Mapping Events to Reactor SLS Files¶

The event tag and data are both critical when working with the reactor system. In the master configuration file under the reactor option, tags are associated with lists of reactor sls formulas (globs can be used for matching):

reactor:
  - 'auth':
    - /srv/reactor/authreact1.sls
    - /srv/reactor/authreact2.sls
  - 'minion_start':
    - /srv/reactor/start.sls

When an event with a tag of auth is fired, the reactor will catch the event and render the two listed files. The rendered files are standard sls files, so by default they are yaml + Jinja. The Jinja is packed with a few data structures similar to state and pillar sls files. The data available is in tag and data variables. The tag variable is just the tag in the fired event and the data variable is the event's data dict. Here is a simple reactor sls:

{% if data['id'] == 'mysql1' %}
highstate_run:
  cmd.state.highstate:
    - tgt: mysql1
{% endif %}

This simple reactor file uses Jinja to further refine the reaction to be made. If the id in the event data is mysql1 (in other words, if the name of the minion is mysql1) then the following reaction is defined. The same data structure and compiler used for the state system is used for the reactor system. The only difference is that the data is matched up to the salt command API and the runner system. In this example, a command is published to the mysql1 minion with a function of state.highstate. Similarly, a runner can be called:

{% if data['data']['overstate'] == 'refresh' %}
overstate_run:
  runner.state.over
{% endif %}

This example will execute the state.overstate runner and initiate an overstate execution.

Fire an event¶

From a minion, run bellow command

salt-call event.fire_master '{"overstate": "refresh"}' 'foo'

In reactor fomular files that are associated with tag foo, data can be accessed via data['data']. Above command passed a dictionary as data, its overstate key can be accessed via data['data']['overstate']. See salt.modules.event for more information.

Understanding the Structure of Reactor Formulas¶

While the reactor system uses the same data structure as the state system, this data does not translate the same way to operations. In state, formulas information is mapped to the state functions, but in the reactor system, information is mapped to a number of available subsystems on the master. These systems are the LocalClient and the Runners. The state declaration field takes a reference to the function to call in each interface. So to trigger a salt-run call the state declaration field will start with runner, followed by the runner function to call. This means that a call to what would be on the command line salt-run manage.up will be runner.manage.up. An example of this in a reactor formula would look like this:

manage_up:
  runner.manage.up

If the runner takes arguments then they can be specified as well:

overstate_dev_env:
  runner.state.over:
    - env: dev

Executing remote commands maps to the LocalClient interface which is used by the salt command. This interface more specifically maps to the cmd_async method inside of the LocalClient class. This means that the arguments passed are being passed to the cmd_async method, not the remote method. A field starts with cmd to use the LocalClient subsystem. The result is, to execute a remote command, a reactor fomular would look like this:

clean_tmp:
  cmd.cmd.run:
    - tgt: '*'
    - arg:
      - rm -rf /tmp/*

The arg option takes a list of arguments as they would be presented on the command line, so the above declaration is the same as running this salt command:

salt '*' cmd.run 'rm -rf /tmp/*'

Use the expr_form argument to specify a matcher:

clean_tmp:
  cmd.cmd.run:
    - tgt: 'os:Ubuntu'
    - expr_form: grain
    - arg:
      - rm -rf /tmp/*


clean_tmp:
  cmd.cmd.run:
    - tgt: 'G@roles:hbase_master'
    - expr_form: compound
    - arg:
      - rm -rf /tmp/*

Salt Formulas¶

Formulas are pre-written Salt States. They are as open-ended as Salt States themselves and can be used for tasks such as installing a package, configuring and starting a service, setting up users or permissions, and many other common tasks.

Note

Formulas require Salt 0.17 or later.

More accurately, Formulas are not tested on earlier versions of Salt so your mileage may vary.

All Formulas require the grains execution module that shipped with Salt 0.16.4. Earlier Salt versions may copy https://github.com/saltstack/salt/blob/develop/salt/modules/grains.py into the /srv/salt/_modules directory and it will be automatically distributed to all minions.

Some Formula utilize features added in Salt 0.17 and will not work on earlier Salt versions.

All official Salt Formulas are found as separate Git repositories in the "saltstack-formulas" organization on GitHub:

https://github.com/saltstack-formulas

As an example, quickly install and configure the popular memcached server using sane defaults simply by including the memcached-formula repository into an existing Salt States tree.

Installation¶

Each Salt Formula is an individual Git repository designed as a drop-in addition to an existing Salt State tree. Formulas can be installed in the following ways.

Adding a Formula as a GitFS remote¶

One design goal of Salt's GitFS fileserver backend was to facilitate reusable States so this is a quick and natural way to use Formulas.

See also

Setting up GitFS

Add one or more Formula repository URLs as remotes in the gitfs_remotes list in the Salt Master configuration file.
Restart the Salt master.

Adding a Formula directory manually¶

Since Formulas are simply directories they can be copied onto the local file system by using Git to clone the repository or by downloading and expanding a tarball or zip file of the directory.

Clone the repository manually and add a new entry to file_roots pointing to the clone's directory.
Clone the repository manually and then copy or link the Formula directory into file_roots.

Usage¶

Each Formula is intended to be immediately usable with sane defaults without any additional configuration. Many formulas are also configurable by including data in Pillar; see the pillar.example file in each Formula repository for available options.

Including a Formula in an existing State tree¶

Formula may be included in an existing sls file. This is often useful when a state you are writing needs to require or extend a state defined in the formula.

Here is an example of a state that uses the epel-formula in a require declaration which directs Salt to not install the python26 package until after the EPEL repository has also been installed:

include:
  - epel

python26:
  pkg:
    - installed
    - require:
      - pkg: epel

Including a Formula from a Top File¶

Some Formula perform completely standalone installations that are not referenced from other state files. It is usually cleanest to include these Formula directly from a Top File.

For example the easiest way to set up an OpenStack deployment on a single machine is to include the openstack-standalone-formula directly from a top.sls file:

base:
  'myopenstackmaster':
    - openstack

Quickly deploying OpenStack across several dedicated machines could also be done directly from a Top File and may look something like this:

base:
  'controller':
    - openstack.horizon
    - openstack.keystone
  'hyper-*':
    - openstack.nova
    - openstack.glance
  'storage-*':
    - openstack.swift

Configuring Formula using Pillar¶

Salt Formulas are designed to work out of the box with no additional configuration. However, many Formula support additional configuration and customization through Pillar. Examples of available options can be found in a file named pillar.example in the root directory of each Formula repository.

Modifying default Formula behavior¶

Remember that Formula are regular Salt States and can be used with all Salt's normal mechanisms for determining execution order. Formula can be required from other States with require declarations, they can be modified using extend, they can made to watch other states with watch_in, they can be used as templates for other States with use. Don't be shy to read through the source for each Formula!

Reporting problems & making additions¶

Each Formula is a separate repository on GitHub. If you encounter a bug with a Formula please file an issue in the respective repository! Send fixes and additions as a pull request. Add tips and tricks to the repository wiki.

Writing Formulas¶

Each Formula is a separate repository in the saltstack-formulas organization on GitHub.

Note

Get involved creating new Formulas

The best way to create new Formula repositories for now is to create a repository in your own account on GitHub and notify a SaltStack employee when it is ready. We will add you as a collaborator on the saltstack-formulas organization and help you transfer the repository over. Ping a SaltStack employee on IRC (#salt on Freenode) or send an email to the Salt mailing list.

Repository structure¶

A basic Formula repository should have the following layout:

foo-formula
|-- foo/
|   |-- map.jinja
|   |-- init.sls
|   `-- bar.sls
|-- LICENSE
|-- pillar.example
`-- README.rst

`README.rst`¶

The README should detail each available .sls file by explaining what it does, whether it has any dependencies on other formulas, whether it has a target platform, and any other installation or usage instructions or tips.

A sample skeleton for the README.rst file:

foo
===

Install and configure the FOO service.

.. note::

    See the full `Salt Formulas installation and usage instructions
    <http://docs.saltstack.com/topics/conventions/formulas.html>`_.

Available states
----------------

``foo``
    Install the ``foo`` package and enable the service.
``foo.bar``
    Install the ``bar`` package.

`map.jinja`¶

It is useful to have a single source for platform-specific or other parameterized information that can be reused throughout a Formula. See "Configuration and parameterization" below for more information. Such a file should be named map.jinja and live alongside the state files.

The following is an example from the MySQL Formula.

map.jinja:

{% set mysql = salt['grains.filter_by']({
    'Debian': {
        'server': 'mysql-server',
        'client': 'mysql-client',
        'service': 'mysql',
        'config': '/etc/mysql/my.cnf',
    },
    'RedHat': {
        'server': 'mysql-server',
        'client': 'mysql',
        'service': 'mysqld',
        'config': '/etc/my.cnf',
    },
    'Gentoo': {
        'server': 'dev-db/mysql',
        'mysql-client': 'dev-db/mysql',
        'service': 'mysql',
        'config': '/etc/mysql/my.cnf',
    },
}, merge=salt['pillar.get']('mysql:lookup')) %}

Any of the values defined above can be fetched for the current platform in any state file using the following syntax:

{% from "mysql/map.jinja" import mysql with context %}

mysql-server:
  pkg:
    - installed
    - name: {{ mysql.server }}
  service:
    - running
    - name: {{ mysql.service }}
    - require:
      - pkg: mysql-server

mysql-config:
  file:
    - managed
    - name: {{ mysql.config }}
    - source: salt://mysql/conf/my.cnf
    - watch:
      - service: mysql-server

SLS files¶

Each state in a Formula should use sane defaults (as much as is possible) and use Pillar to allow for customization.

The root state, in particular, and most states in general, should strive to do no more than the basic expected thing and advanced configuration should be put in child states build on top of the basic states.

For example, the root Apache should only install the Apache httpd server and make sure the httpd service is running. It can then be used by more advanced states:

# apache/init.sls
httpd:
  pkg:
    - installed
  service:
    - running

# apache/mod_wsgi.sls
include:
  - apache

mod_wsgi:
  pkg:
    - installed
    - require:
      - pkg: apache

# apache/debian/vhost_setup.sls
{% if grains['os_family'] == 'Debian' %}
a2dissite 000-default:
  cmd.run:
    - onlyif: test -L /etc/apache2/sites-enabled/000-default
    - require:
      - pkg: apache
{% endif %}

Platform agnostic¶

Each Salt Formula must be able to be run without error on any platform. If the formula is not applicable to a platform it should do nothing. See the epel-formula for an example.

Any platform-specific states must be wrapped in conditional statements:

{% if grains['os_family'] == 'Debian' %}
...
{% endif %}

A handy method for using platform-specific values is to create a lookup table using the filter_by() function:

{% set apache = salt['grains.filter_by']({
    'Debian': {'conf': '/etc/apache2/conf.d'},
    'RedHat': {'conf': '/etc/httpd/conf.d'},
}) %}

myconf:
  file:
    - managed
    - name: {{ apache.conf }}/myconf.conf

Configuration and parameterization¶

Each Formula should strive for sane defaults that can then be customized using Pillar. Pillar lookups must use the safe get() and must provide a default value:

{% if salt['pillar.get']('horizon:use_ssl', False) %}
ssl_crt: {{ salt['pillar.get']('horizon:ssl_crt', '/etc/ssl/certs/horizon.crt') }}
ssl_key: {{ salt['pillar.get']('horizon:ssl_key', '/etc/ssl/certs/horizon.key') }}
{% endif %}

Any default values used in the Formula must also be documented in the pillar.example file in the root of the repository. Comments should be used liberally to explain the intent of each configuration value. In addition, users should be able copy-and-paste the contents of this file into their own Pillar to make any desired changes.

Scripting¶

Remember that both State files and Pillar files can easily call out to Salt execution modules and have access to all the system grains as well.

{% if '/storage' in salt['mount.active']() %}
/usr/local/etc/myfile.conf:
  file:
    - symlink
    - target: /storage/myfile.conf
{% endif %}

Jinja macros are generally discouraged in favor of adding functions to existing Salt modules or adding new modules. An example of this is the filter_by() function.

Versioning¶

Formula versions are tracked using Git tags.

Testing Formulas¶

Salt Formulas are tested by running each .sls file via state.sls and checking the output for success or failure. This is done for each supported platform.

Salt Conventions¶

Salt Release Process¶

The goal for Salt projects is to cut a new feature release every four to six weeks. This document outlines the process for these releases, and the subsequent bug fix releases which follow.

Feature Release Process¶

When a new release is ready to be cut, the person responsible for cutting the release will follow the following steps (written using the 0.16 release as an example):

All open issues on the release milestone should be moved to the next release milestone. (e.g. from the 0.16 milestone to the 0.17 milestone)
Release notes should be created documenting the major new features and bugfixes in the release.
Create an annotated tag with only the major and minor version numbers, preceded by the letter v. (e.g. v0.16) This tag will reside on the develop branch.
Create a branch for the new release, using only the major and minor version numbers. (e.g. 0.16)
On this new branch, create an annotated tag for the first revision release, which is generally a release candidate. It should be preceded by the letter v. (e.g. v0.16.0RC)
The release should be packaged from this annotated tag and uploaded to PyPI.
The packagers should be notified on the salt-packagers mailing list so they can create packages for all the major operating systems. (note that release candidates should go in the testing repositories)
After the packagers have been given a few days to compile the packages, the release is announced on the salt-users mailing list.
Log into RTD and add the new release there. (Have to do it manually)

Maintenance and Bugfix Releases¶

Once a release has been cut, regular cherry-picking sessions should begin to cherry-pick any bugfixes from the develop branch to the release branch (e.g. 0.16). Once major bugs have been fixes and cherry-picked, a bugfix release can be cut:

On the release branch (i.e. 0.16), create an annotated tag for the revision release. It should be preceded by the letter v. (e.g. v0.16.2) Release candidates are unnecessary for bugfix releases.
The release should be packaged from this annotated tag and uploaded to PyPI.
The packagers should be notified on the salt-packagers mailing list so they can create packages for all the major operating systems.
After the packagers have been given a few days to compile the packages, the release is announced on the salt-users mailing list.

Salt Coding Style¶

Salt is developed with a certain coding style, while the style is dominantly PEP 8 it is not completely PEP 8. It is also noteworthy that a few development techniques are also employed which should be adhered to. In the end, the code is made to be "Salty".

Most importantly though, we will accept code that violates the coding style and KINDLY ask the contributor to fix it, or go ahead and fix the code on behalf of the contributor. Coding style is NEVER grounds to reject code contributions, and is never grounds to talk down to another member of the community (There are no grounds to treat others without respect, especially people working to improve Salt)!!

Strings¶

Salt follows a few rules when formatting strings:

Single Quotes¶

In Salt, all strings use single quotes unless there is a good reason not to. This means that docstrings use single quotes, standard strings use single quotes etc.:

def foo():
    '''
    A function that does things
    '''
    name = 'A name'
    return name

Formatting Strings¶

All strings which require formatting should use the .format string method:

data = 'some text'
more = '{0} and then some'.format(data)

Make sure to use indices or identifiers in the format brackets, since empty brackets are not supported by python 2.6.

Please do NOT use printf formatting.

Docstring Conventions¶

Docstrings should always add a newline, docutils takes care of the new line and it makes the code cleaner and more vertical:

GOOD:

def bar():
    '''
    Here lies a docstring with a newline after the quotes and is the salty
    way to handle it! Vertical code is the way to go!
    '''
    return

BAD:

def baz():
    '''This is not ok!'''
    return

Imports¶

Salt code prefers importing modules and not explicit functions. This is both a style and functional preference. The functional preference originates around the fact that the module import system used by pluggable modules will include callable objects (functions) that exist in the direct module namespace. This is not only messy, but may unintentionally expose code python libs to the Salt interface and pose a security problem.

To say this more directly with an example, this is GOOD:

import os

def minion_path():
    path = os.path.join(self.opts['cachedir'], 'minions')
    return path

This on the other hand is DISCOURAGED:

from os.path import join

def minion_path():
    path = join(self.opts['cachedir'], 'minions')
    return path

The time when this is changed is for importing exceptions, generally directly importing exceptions is preferred:

This is a good way to import exceptions:

from salt.exceptions import CommandExecutionError

Absolute Imports¶

Although absolute imports seems like an awesome idea, please do not use it. Extra care would be necessary all over salt's code in order for absolute imports to work as supposed. Believe it, it has been tried before and, as a tried example, by renaming salt.modules.sysmod to salt.modules.sys, all other salt modules which needed to import sys would have to also import absolute_import, which should be avoided.

Vertical is Better¶

When writing Salt code, vertical code is generally preferred. This is not a hard rule but more of a guideline. As PEP 8 specifies, Salt code should not exceed 79 characters on a line, but it is preferred to separate code out into more newlines in some cases for better readability:

import os

os.chmod(
        os.path.join(self.opts['sock_dir'],
            'minion_event_pub.ipc'),
        448
        )

Where there are more line breaks, this is also apparent when constructing a function with many arguments, something very common in state functions for instance:

def managed(name,
        source=None,
        source_hash='',
        user=None,
        group=None,
        mode=None,
        template=None,
        makedirs=False,
        context=None,
        replace=True,
        defaults=None,
        env=None,
        backup='',
        **kwargs):

Note

Making function and class definitions vertical is only required if the arguments are longer then 80 characters. Otherwise, the formatting is optional and both are acceptable.

Indenting¶

Some confusion exists in the python world about indenting things like function calls, the above examples use 8 spaces when indenting comma-delimited constructs.

The confusion arises because the pep8 program INCORRECTLY flags this as wrong, where PEP 8, the document, cites only using 4 spaces here as wrong, as it doesn't differentiate from a new indent level.

Right:

def managed(name,
        source=None,
        source_hash='',
        user=None)

WRONG:

def managed(name,
    source=None,
    source_hash='',
    user=None)

Lining up the indent is also correct:

def managed(name,
            source=None,
            source_hash='',
            user=None)

This also applies to function calls and other hanging indents.

pep8 and Flake8 (and, by extension, the vim plugin Syntastic) will complain about the double indent for hanging indents. This is a known conflict between pep8 (the script) and the actual PEP 8 standard. It is recommended that this particular warning be ignored with the following lines in ~/.config/flake8:

[flake8]
ignore = E226,E241,E242,E126

Make sure your Flake8/pep8 are up to date. The first three errors are ignored by default and are present here to keep the behavior the same. This will also work for pep8 without the Flake8 wrapper -- just replace all instances of 'flake8' with 'pep8', including the filename.

Code Churn¶

Many pull requests have been submitted that only churn code in the name of PEP 8. Code churn is a leading source of bugs and is strongly discouraged. While style fixes are encouraged they should be isolated to a single file per commit, and the changes should be legitimate, if there are any questions about whether a style change is legitimate please reference this document and the official PEP 8 (http://www.python.org/dev/peps/pep-0008/) document before changing code. Many claims that a change is PEP 8 have been invalid, please double check before committing fixes.

Salt Stack Git Policy¶

The Salt Stack team follows a git policy to maintain stability and consistency with the repository. The git policy has been developed to encourage contributions and make contributing to Salt as easy as possible. Code contributors to Salt Stack projects DO NOT NEED TO READ THIS DOCUMENT, because all contributions come into Salt Stack via a single gateway to make it as easy as possible for contributors to give us code.

The primary rule of git management in Salt Stack is to make life easy on contributors and developers to send in code. Simplicity is always a goal!

New Code Entry¶

All new Salt Stack code is posted to the develop branch, this is the single point of entry. The only exception here is when a bugfix to develop cannot be cleanly merged into a release branch and the bugfix needs to be rewritten for the release branch.

Release Branching¶

Salt Stack maintains two types of releases, Feature Releases and Point Releases. A feature release is managed by incrementing the first or second release point number, so 0.10.5 -> 0.11.0 signifies a feature release and 0.11.0 -> 0.11.1 signifies a point release, also a hypothetical 0.42.7 -> 1.0.0 would also signify a feature release.

Feature Release Branching¶

Each feature release is maintained in a dedicated git branch derived from the last applicable release commit on develop. All file changes relevant to the feature release will be completed in the develop branch prior to the creation of the feature release branch. The feature release branch will be named after the relevant numbers to the feature release, which constitute the first two numbers. This means that the release branch for the 0.11.0 series is named 0.11.

A feature release branch is created with the following command:

# git checkout -b 0.11 # From the develop branch
# git push origin 0.11

Point Releases¶

Each point release is derived from its parent release branch. Constructing point releases is a critical aspect of Salt development and is managed by members of the core development team. Point releases comprise bug and security fixes which are cherry picked from develop onto the aforementioned release branch. At the time when a core developer accepts a pull request a determination needs to be made if the commits in the pull request need to be backported to the release branch. Some simple criteria are used to make this determination:

Is this commit fixing a bug? Backport
Does this commit change or add new features in any way? Don't backport
Is this a PEP8 or code cleanup commit? Don't backport
Does this commit fix a security issue? Backport

Determining when a point release is going to be made is up to the project leader (Thomas Hatch). Generally point releases are made every 1-2 weeks or if there is a security fix they can be made sooner.

The point release is only designated by tagging the commit on the release branch with release number using the existing convention (version 0.11.1 is tagged with v0.11.1). From the tag point a new source tarball is generated and published to PyPI, and a release announcement is made.

Salt Development Guidelines¶

Deprecating Code¶

Salt should remain backwards compatible, though sometimes, this backwards compatibility needs to be broken because a specific feature and/or solution is no longer necessary or required. At first one might think, let me change this code, it seems that it's not used anywhere else so it should be safe to remove. Then, once there's a new release, users complain about functionality which was removed and they where using it, etc. This should, at all costs, be avoided, and, in these cases, that specific code should be deprecated.

Depending on the complexity and usage of a specific piece of code, the deprecation time frame should be properly evaluated. As an example, a deprecation warning which is shown for 2 major releases, for example 0.17.0 and 0.18.0, gives users enough time to stop using the deprecated code and adapt to the new one.

For example, if you're deprecating the usage of a keyword argument to a function, that specific keyword argument should remain in place for the full deprecation time frame and if that keyword argument is used, a deprecation warning should be shown to the user.

To help in this deprecation task, salt provides salt.utils.warn_until. The idea behind this helper function is to show the deprecation warning until salt reaches the provided version. Once that provided version is equaled salt.utils.warn_until will raise a RuntimeError making salt stop its execution. This stoppage is unpleasant and will remind the developer that the deprecation limit has been reached and that the code can then be safely removed.

Consider the following example:

def some_function(bar=False, foo=None):
    if foo is not None:
        salt.utils.warn_until(
            (0, 18),
            'The \'foo\' argument has been deprecated and its '
            'functionality removed, as such, its usage is no longer '
            'required.'
        )

Consider that the current salt release is 0.16.0. Whenever foo is passed a value different from None that warning will be shown to the user. This will happen in versions 0.16.2 to 0.18.0, after which a RuntimeError will be raised making us aware that the deprecated code should now be removed.

Dunder Dictionaries¶

Salt provides several special "dunder" dictionaries as a convenience for Salt development. These include __opts__, __context__, __salt__, and others. This document will describe each dictionary and detail where they exist and what information and/or functionality they provide.

context¶

__context__ exists in state modules and execution modules.

During a state run the __context__ dictionary persists across all states that are run and then is destroyed when the state ends.

When running an execution module __context__ persists across all module executions until the modules are refreshed; such as when saltutils.sync_all or state.highstate are executed.

A great place to see how to use __context__ is in the cp.py module in salt/modules/cp.py. The fileclient authenticates with the master when it is instantiated and then is used to copy files to the minion. Rather than create a new fileclient for each file that is to be copied down, one instance of the fileclient is instantiated in the __context__ dictionary and is reused for each file. Here is an example from salt/modules/cp.py:

if not 'cp.fileclient' in __context__:
    __context__['cp.fileclient'] = salt.fileclient.get_file_client(__opts__)

Note

Because __context__ may or may not have been destroyed, always be sure to check for the existence of the key in __context__ and generate the key before using it.

External Pillars¶

Salt provides a mechanism for generating pillar data by calling external pillar interfaces. This document will describe an outline of an ext_pillar module.

Location¶

Salt expects to find your ext_pillar module in the same location where it looks for other python modules. If the extension_modules option in your Salt master configuration is set, Salt will look for a pillar directory under there and load all the modules it finds. Otherwise, it will look in your Python site-packages salt/pillar directory.

Configuration¶

The external pillars that are called when a minion refreshes its pillars is controlled by the ext_pillar option in the Salt master configuration. You can pass a single argument, a list of arguments or a dictionary of arguments to your pillar:

ext_pillar:
  - example_a: some argument
  - example_b:
    - argumentA
    - argumentB
  - example_c:
      keyA: valueA
      keyB: valueB

The Module¶

Imports and Logging¶

Import modules your external pillar module needs. You should first include generic modules that come with stock Python:

import logging

And then start logging. This is an idiomatic way of setting up logging in Salt:

log = logging.getLogger(__name__)

Finally, load modules that are specific to what you are doing. You should catch import errors and set a flag that the the __virtual__ function can use later.

try:
    import weird_thing
    example_a_loaded = True
except ImportError:
    example_a_loaded = False

Options¶

If you define an __opts__ dictionary, it will be merged into the __opts__ dictionary handed to the ext_pillar function later. This is a good place to put default configuration items. The convention is to name things modulename.option.

__opts__ = { 'example_a.someconfig': 137 }

Initialization¶

If you define an __init__ function, it will be called with the following signature:

def __init__( __opts__ ):
    # Do init work here

Note: The __init__ function is ran every time a particular minion causes the external pillar to be called, so don't put heavy initialization code here. The __init__ functionality is a side-effect of the Salt loader, so it may not be as useful in pillars as it is in other Salt items.

virtual¶

If you define a __virtual__ function, you can control whether or not this module is visible. If it returns False then Salt ignores this module. If it returns a string, then that string will be how Salt identifies this external pillar in its ext_pillar configuration. If this function does not exist, then the name Salt's ext_pillar will use to identify this module is its conventional name in Python.

This is useful to write modules that can be installed on all Salt masters, but will only be visible if a particular piece of software your module requires is installed.

# This external pillar will be known as `example_a`
def __virtual__():
    if example_a_loaded:
        return 'example_a'
    else:
        return False

# This external pillar will be known as `something_else`
def __virtual__():
    if example_a_loaded:
        return 'something_else'
    else:
        return False

ext_pillar¶

This is where the real work of an external pillar is done. If this module is active and has a function called ext_pillar, whenever a minion updates its pillar this function is called.

How it is called depends on how it is configured in the Salt master configuration. The first argument is always the current pillar dictionary, this contains pillar items that have already been added, starting with the data from pillar_roots, and then from any already-ran external pillars.

Using our example above:

ext_pillar( pillar, 'some argument' )                   # example_a
ext_pillar( pillar, 'argumentA', 'argumentB' )          # example_b
ext_pillar( pillar, keyA='valueA', keyB='valueB' } )    # example_c

In the example_a case, pillar will contain the items from the pillar_roots, in example_b pillar will contain that plus the items added by example_a, and in example_c pillar will contain that plus the items added by example_b.

This function should return a dictionary, the contents of which are merged in with all of the other pillars and returned to the minion. Note: this function is called once for each minion that fetches its pillar data.

def ext_pillar( pillar, *args, **kwargs ):

    my_pillar = {}

    # Do stuff

    return my_pillar

You shouldn't just add items to pillar and return that, since that will cause Salt to merge data that already exists. Rather, just return the items you are adding or changing. You could, however, use pillar in your module to make some decision based on pillar data that already exists.

This function has access to some useful globals:

__opts__:	A dictionary of mostly Salt configuration options. If you had an `__opts__` dictionary defined in your module, those values will be included. Also included and most useful is `__opts__['id']`, which is the minion id of the minion asking for pillar data.
__salt__:	A dictionary of Salt module functions, useful so you don't have to duplicate functions that already exist. E.g. `__salt__['cmd.run']( 'ls -l' )` Note, runs on the master
__grains__:	A dictionary of the grains of the minion making this pillar call.

Example configuration¶

As an example, if you wanted to add external pillar via the cmd_json external pillar, add something like this to your master config:

ext_pillar:
  - cmd_json: "echo {'arg':'value'}"

Logging Internals¶

TODO

Modular Systems¶

When first working with Salt, it is not always clear where all of the modular components are and what they do. Salt comes loaded with more modular systems than many users are aware of, making Salt very easy to extend in many places.

The most commonly used modular systems are execution modules and states. But the modular systems extend well beyond the more easily exposed components and are often added to Salt to make the complete system more flexible.

Execution Modules¶

Execution modules make up the core of the functionality used by Salt to interact with client systems. The execution modules create the core system management library used by all Salt systems, including states, which interact with minion systems.

Execution modules are completely open ended in their execution. They can be used to do anything required on a minion, from installing packages to detecting information about the system. The only restraint in execution modules is that the defined functions always return a JSON serializable object.

For a list of all built in execution modules, click here

For information on writing execution modules, see this page.

State Modules¶

State modules are used to define the state interfaces used by Salt States. These modules are restrictive in that they must follow a number of rules to function properly.

Note

State modules define the available routines in sls files. If calling an execution module directly is desired, take a look at the module state.

Auth¶

The auth module system allows for external authentication routines to be easily added into Salt. The auth function needs to be implemented to satisfy the requirements of an auth module. Use the pam module as an example.

Fileserver¶

The fileserver module system is used to create fileserver backends used by the Salt Master. These modules need to implement the functions used in the fileserver subsystem. Use the gitfs module as an example.

Grains¶

Grain modules define extra routines to populate grains data. All defined public functions will be executed and MUST return a Python dict object. The dict keys will be added to the grains made available to the minion.

Output¶

The output modules supply the outputter system with routines to display data in the terminal. These modules are very simple and only require the output function to execute. The default system outputter is the nested module.

Pillar¶

Used to define optional external pillar systems. The pillar generated via the filesystem pillar is passed into external pillars. This is commonly used as a bridge to database data for pillar, but is also the backend to the libvirt state used to generate and sign libvirt certificates on the fly.

Renderers¶

Renderers are the system used to render sls files into salt highdata for the state compiler. They can be as simple as the py renderer and as complex as stateconf and pydsl.

Returners¶

Returners are used to send data from minions to external sources, commonly databases. A full returner will implement all routines to be supported as an external job cache. Use the redis returner as an example.

Runners¶

Runners are purely master-side execution sequences. These range from simple reporting to orchestration engines like the overstate.

Tops¶

Tops modules are used to convert external data sources into top file data for the state system.

Wheel¶

The wheel system is used to manage master side management routines. These routines are primarily intended for the API to enable master configuration.

Package Providers¶

This page contains guidelines for writing package providers.

Package Functions¶

One of the most important features of Salt is package management. There is no shortage of package managers, so in the interest of providing a consistent experience in pkg states, there are certain functions that should be present in a package provider. Note that these are subject to change as new features are added or existing features are enhanced.

list_pkgs¶

This function should declare an empty dict, and then add packages to it by calling pkg_resource.add_pkg, like so:

__salt__['pkg_resource.add_pkg'](ret, name, version)

The last thing that should be done before returning is to execute pkg_resource.sort_pkglist. This function does not presently do anything to the return dict, but will be used in future versions of Salt.

__salt__['pkg_resource.sort_pkglist'](ret)

list_pkgs returns a dictionary of installed packages, with the keys being the package names and the values being the version installed. Example return data:

{'foo': '1.2.3-4',
 'bar': '5.6.7-8'}

latest_version¶

Accepts an arbitrary number of arguments. Each argument is a package name. The return value for a package will be an empty string if the package is not found or if the package is up-to-date. The only case in which a non-empty string is returned is if the package is available for new installation (i.e. not already installed) or if there is an upgrade available.

If only one argument was passed, this function return a string, otherwise a dict of name/version pairs is returned.

This function must also accept **kwargs, in order to receive the fromrepo and repo keyword arguments from pkg states. Where supported, these arguments should be used to find the install/upgrade candidate in the specified repository. The fromrepo kwarg takes precedence over repo, so if both of those kwargs are present, the repository specified in fromrepo should be used. However, if repo is used instead of fromrepo, it should still work, to preserve backwards compatibility with older versions of Salt.

version¶

Like latest_version, accepts an arbitrary number of arguments and returns a string if a single package name was passed, or a dict of name/value pairs if more than one was passed. The only difference is that the return values are the currently-installed versions of whatever packages are passed. If the package is not installed, an empty string is returned for that package.

upgrade_available¶

Deprecated and destined to be removed. For now, should just do the following:

return __salt__['pkg.latest_version'](name) != ''

install¶

The following arguments are required and should default to None:

name (for single-package pkg states)
pkgs (for multiple-package pkg states)
sources (for binary package file installation)

The first thing that this function should do is call pkg_resource.parse_targets (see below). This function will convert the SLS input into a more easily parsed data structure. pkg_resource.parse_targets may need to be modified to support your new package provider, as it does things like parsing package metadata which cannot be done for every package management system.

pkg_params, pkg_type = __salt__['pkg_resource.parse_targets'](name,
                                                              pkgs,
                                                              sources)

Two values will be returned to the install function. The first of them will be a dictionary. The keys of this dictionary will be package names, though the values will differ depending on what kind of installation is being done:

If name was provided (and pkgs was not), then there will be a single key in the dictionary, and its value will be None. Once the data has been returned, if the version keyword argument was provided, then it should replace the None value in the dictionary.
If pkgs was provided, then name is ignored, and the dictionary will contain one entry for each package in the pkgs list. The values in the dictionary will be None if a version was not specified for the package, and the desired version if specified. See the Multiple Package Installation Options section of the pkg.installed state for more info.
If sources was provided, then name is ignored, and the dictionary values will be the path/URI for the package.

The second return value will be a string with two possible values: repository or file. The install function can use this value (if necessary) to build the proper command to install the targeted package(s).

Both before and after the installing the target(s), you should run list_pkgs to obtain a list of the installed packages. You should then return the output of pkg_resource.find_changes:

return __salt__['pkg_resource.find_changes'](old, new)

remove¶

Removes the passed package and return a list of the packages removed.

Package Repo Functions¶

There are some functions provided by pkg which are specific to package repositories, and not to packages themselves. When writing modules for new package managers, these functions should be made available as stated below, in order to provide compatibility with the pkgrepo state.

All repo functions should accept a basedir option, which defines which directory repository configuration should be found in. The default for this is dictated by the repo manager that is being used, and rarely needs to be changed.

basedir = '/etc/yum.repos.d'
__salt__['pkg.list_repos'](basedir)

list_repos¶

Lists the repositories that are currently configured on this system.

__salt__['pkg.list_repos']()

Returns a dictionary, in the following format:

{'reponame': 'config_key_1': 'config value 1',
             'config_key_2': 'config value 2',
             'config_key_3': ['list item 1 (when appropriate)',
                              'list item 2 (when appropriate)]}

get_repo¶

Displays all local configuration for a specific repository.

__salt__['pkg.get_repo'](repo='myrepo')

The information is formatted in much the same way as list_repos, but is specific to only one repo.

{'config_key_1': 'config value 1',
 'config_key_2': 'config value 2',
 'config_key_3': ['list item 1 (when appropriate)',
                  'list item 2 (when appropriate)]}

del_repo¶

Removes the local configuration for a specific repository. Requires a repo argument, which must match the locally configured name. This function returns a string, which informs the user as to whether or not the operation was a success.

__salt__['pkg.del_repo'](repo='myrepo')

mod_repo¶

Modify the local configuration for one or more option for a configured repo. This is also the way to create new repository configuration on the local system; if a repo is specified which does not yet exist, it will be created.

The options specified for this function are specific to the system; please refer to the documentation for your specific repo manager for specifics.

__salt__['pkg.mod_repo'](repo='myrepo', url='http://myurl.com/repo')

Low-Package Functions¶

In general, the standard package functions as describes above will meet your needs. These functions use the system's native repo manager (for instance, yum or the apt tools). In most cases, the repo manager is actually separate from the package manager. For instance, yum is usually a front-end for rpm, and apt is usually a front-end for dpkg. When possible, the package functions that use those package managers directly should do so through the low package functions.

It is normal and sane for pkg to make calls to lowpkgs, but lowpkg must never make calls to pkg. This is affects functions which are required by both pkg and lowpkg, but the technique in pkg is more performant than what is available to lowpkg. When this is the case, the lowpkg function that requires that technique must still use the lowpkg version.

list_pkgs¶

Returns a dict of packages installed, including the package name and version. Can accept a list of packages; if none are specified, then all installed packages will be listed.

installed = __salt__['lowpkg.list_pkgs']('foo', 'bar')

Example output:

{'foo': '1.2.3-4',
 'bar': '5.6.7-8'}

verify¶

Many (but not all) package management systems provide a way to verify that the files installed by the package manager have or have not changed. This function accepts a list of packages; if none are specified, all packages will be included.

installed = __salt__['lowpkg.verify']('httpd')

Example output:

{'/etc/httpd/conf/httpd.conf': {'mismatch': ['size', 'md5sum', 'mtime'],
                                'type': 'config'}}

file_list¶

Lists all of the files installed by all packages specified. If not packages are specified, then all files for all known packages are returned.

installed = __salt__['lowpkg.file_list']('httpd', 'apache')

This function does not return which files belong to which packages; all files are returned as one giant list (hence the file_list function name. However, This information is still returned inside of a dict, so that it can provide any errors to the user in a sane manner.

{'errors': ['package apache is not installed'],
  'files': ['/etc/httpd',
            '/etc/httpd/conf',
            '/etc/httpd/conf.d',
            '...SNIP...']}

file_dict¶

Lists all of the files installed by all packages specified. If not packages are specified, then all files for all known packages are returned.

installed = __salt__['lowpkg.file_dict']('httpd', 'apache', 'kernel')

Unlike file_list, this function will break down which files belong to which packages. It will also return errors in the same manner as file_list.

{'errors': ['package apache is not installed'],
 'packages': {'httpd': ['/etc/httpd',
                        '/etc/httpd/conf',
                        '...SNIP...'],
              'kernel': ['/boot/.vmlinuz-2.6.32-279.el6.x86_64.hmac',
                         '/boot/System.map-2.6.32-279.el6.x86_64',
                         '...SNIP...']}}

Logging¶

The salt project tries to get the logging to work for you and help us solve any issues you might find along the way.

If you want to get some more information on the nitty-gritty of salt's logging system, please head over to the logging development document, if all you're after is salt's logging configurations, please continue reading.

Available Configuration Settings¶

`log_file`¶

The log records can be sent to a regular file, local path name, or network location. Remote logging works best when configured to use rsyslogd(8) (e.g.: file:///dev/log), with rsyslogd(8) configured for network logging. The format for remote addresses is: <file|udp|tcp>://<host|socketpath>:<port-if-required>/<log-facility>.

Default: Dependent of the binary being executed, for example, for salt-master, /var/log/salt/master.

Examples:

log_file: /var/log/salt/master

log_file: /var/log/salt/minion

log_file: file:///dev/log

log_file: udp://loghost:10514

`log_level`¶

Default: warning

The level of log record messages to send to the console. One of all, garbage, trace, debug, info, warning, error, critical, quiet.

log_level: warning

`log_level_logfile`¶

Default: warning

The level of messages to send to the log file. One of all, garbage, trace, debug, info, warning, error, critical, quiet.

log_level_logfile: warning

`log_datefmt`¶

Default: %H:%M:%S

The date and time format used in console log messages. Allowed date/time formatting can be seen on time.strftime.

log_datefmt: '%H:%M:%S'

`log_datefmt_logfile`¶

Default: %Y-%m-%d %H:%M:%S

The date and time format used in log file messages. Allowed date/time formatting can be seen on time.strftime.

log_datefmt_logfile: '%Y-%m-%d %H:%M:%S'

`log_fmt_console`¶

Default: [%(levelname)-8s] %(message)s

The format of the console logging messages. Allowed formatting options can be seen on the LogRecord attributes.

log_fmt_console: '[%(levelname)-8s] %(message)s'

`log_fmt_logfile`¶

Default: %(asctime)s,%(msecs)03.0f [%(name)-17s][%(levelname)-8s] %(message)s

The format of the log file logging messages. Allowed formatting options can be seen on the LogRecord attributes.

log_fmt_logfile: '%(asctime)s,%(msecs)03.0f [%(name)-17s][%(levelname)-8s] %(message)s'

`log_granular_levels`¶

Default: {}

This can be used to control logging levels more specifically. The example sets the main salt library at the 'warning' level, but sets salt.modules to log at the debug level:

log_granular_levels:
  'salt': 'warning',
  'salt.modules': 'debug'

External Logging Handlers¶

Besides the internal logging handlers used by salt, there are some external which can be used, see the external logging handlers document.

External Logging Handlers¶

logstash_mod

codeauthor:	Pedro Algarvio (pedro@algarvio.me)

sentry_mod

codeauthor:	Pedro Algarvio (pedro@algarvio.me)

Introduction to Extending Salt¶

Salt is made to be used, and made to be extended. The primary goal of Salt is to provide a foundation which can be used to solve problems. And the goal of Salt is to not assume what those problems might be.

One of the greatest benefit of developing Salt has been the vast array of ways in which people have wanted to use it, while the original intention was as a communication layer for a cloud controller Salt has been extended to facilitate so much more.

Client API¶

The primary interface used to extend Salt, is to simply use it. Salt executions can be called via the Salt client API, making programming master side solutions with Salt is easy.

See also

Python client API

Adding Loadable Plugins¶

Salt is comprised of a core platform that loads many types of easy to write plugins. The idea is to enable all of the breaking points in the Salt processes to have a point of pluggable interaction. This means that all of the main features of Salt can be extended, modified or used.

The breaking points and helping interfaces span from convenience master side executions to manipulating the flow of how data is handled by Salt.

Minion Execution Modules¶

The minion execution modules or just modules are the core to what Salt is and does. These modules are found in:

https://github.com/saltstack/salt/blob/develop/salt/modules

These modules are what is called by the Salt command line and the salt client API. Adding modules is done by simply adding additional Python modules to the modules directory and restarting the minion.

Grains¶

Salt grains, or "grains of truth" are bits of static information that are generated when the minion starts. This information is useful when determining what package manager to default to, or where certain configuration files are stored on the minion.

The Salt grains are the interface used for auto detection and dynamic assignment of execution modules and types to specific Salt minions.

The code used to generate the Salt grains can be found here:

https://github.com/saltstack/salt/blob/develop/salt/grains

States¶

Salt supports state enforcement, this makes Salt a high speed and very efficient solution for system configuration management.

States can be easily added to Salt by dropping a new state module in:

https://github.com/saltstack/salt/blob/develop/salt/states

Renderers¶

Salt states are controlled by simple data structures, these structures can be abstracted in a number of ways. While the default is to be in a YAML file wrapped in a jinja template, any abstraction can be used. This means that any format that can be dreamed is possible, so long as a renderer is written for it.

The existing renderers can be found here:

https://github.com/saltstack/salt/blob/develop/salt/renderers

Returners¶

The Salt commands all produce a return value, that return value is sent to the Salt master by default, but it can be sent anywhere. The returner interface makes it programmatically possible for the information to be sent to anything from an SQL or NoSQL database, to a custom application made to use Salt.

The existing returners can be found here:

https://github.com/saltstack/salt/blob/develop/salt/returners

Runners¶

Sometimes a certain application can be made to execute and run from the existing Salt command line. This is where the Salt runners come into play. The Salt Runners what is called by the Salt-run command and are meant to act as a generic interface for encapsulating master side executions.

Existing Salt runners are located here:

https://github.com/saltstack/salt/blob/develop/salt/runners

Modules¶

Salt modules are the functions called by the salt command.

See also

Full list of builtin modules

Salt ships with many modules that cover a wide variety of tasks.

Modules Are Easy to Write!¶

Salt modules are amazingly simple to write. Just write a regular Python module or a regular Cython module and place it a directory called _modules/ within the file_roots specified by the master config file, and they will be synced to the minions when state.highstate is run, or by executing the saltutil.sync_modules or saltutil.sync_all functions.

Any custom modules which have been synced to a minion, that are named the same as one of Salt's default set of modules, will take the place of the default module with the same name. Note that a module's default name is its filename (i.e. foo.py becomes module foo), but that its name can be overridden by using a __virtual__ function.

Since Salt modules are just Python/Cython modules, there are no restraints on what you can put inside of a Salt module. If a Salt module has errors and cannot be imported, the Salt minion will continue to load without issue and the module with errors will simply be omitted.

If adding a Cython module the file must be named <modulename>.pyx so that the loader knows that the module needs to be imported as a Cython module. The compilation of the Cython module is automatic and happens when the minion starts, so only the *.pyx file is required.

Cross Calling Modules¶

All of the Salt modules are available to each other, and can be "cross called". This means that, when creating a module, functions in modules that already exist can be called.

The variable __salt__ is packed into the modules after they are loaded into the Salt minion. This variable is a Python dictionary of all of the Salt functions, laid out in the same way that they are made available to the Salt command.

Salt modules can be cross called by accessing the value in the __salt__ dict:

def foo(bar):
    return __salt__['cmd.run'](bar)

This code will call the Salt cmd module's run function and pass the argument bar.

Preloaded Modules Data¶

When interacting with modules often it is nice to be able to read information dynamically about the minion, or load in configuration parameters for a module. Salt allows for different types of data to be loaded into the modules by the minion, as of this writing Salt loads information gathered from the Salt Grains system and from the minion configuration file.

Grains Data¶

The Salt minion detects information about the system when started. This allows for modules to be written dynamically with respect to the underlying hardware and operating system. This information is referred to as Salt Grains, or "grains of salt". The Grains system was introduced to replace Facter, since relying on a Ruby application from a Python application was both slow and inefficient. Grains support replaces Facter in all Salt releases after 0.8

The values detected by the Salt Grains on the minion are available in a dict named __grains__ and can be accessed from within callable objects in the Python modules.

To see the contents of the grains dict for a given system in your deployment run the grains.items() function:

salt 'hostname' grains.items

To use the __grains__ dict simply call it as a Python dict from within your code, an excellent example is available in the Grains module: salt.modules.grains.

Module Configuration¶

Since parameters for configuring a module may be desired, Salt allows for configuration information stored in the main minion config file to be passed to the modules.

Since the minion configuration file is a YAML document, arbitrary configuration data can be passed in the minion config that is read by the modules. It is strongly recommended that the values passed in the configuration file match the module. This means that a value intended for the test module should be named test.<value>.

Configuration also requires that default configuration parameters need to be loaded as well. This can be done simply by adding the __opts__ dict to the top level of the module.

The test module contains usage of the module configuration, and the default configuration file for the minion contains the information and format used to pass data to the modules. salt.modules.test, conf/minion.

Printout Configuration¶

Since module functions can return different data, and the way the data is printed can greatly change the presentation, Salt has a printout configuration.

When writing a module the __outputter__ dict can be declared in the module. The __outputter__ dict contains a mapping of function name to Salt Outputter.

__outputter__ = {
                'run': 'txt'
                }

This will ensure that the text outputter is used.

Virtual Modules¶

Sometimes a module should be presented in a generic way. A good example of this can be found in the package manager modules. The package manager changes from one operating system to another, but the Salt module that interfaces with the package manager can be presented in a generic way.

The Salt modules for package managers all contain a __virtual__ function which is called to define what systems the module should be loaded on.

The __virtual__ function is used to return either a string or False. If False is returned then the module is not loaded, if a string is returned then the module is loaded with the name of the string.

This means that the package manager modules can be presented as the pkg module regardless of what the actual module is named.

The package manager modules are the best example of using the __virtual__ function: https://github.com/saltstack/salt/blob/develop/salt/modules/pacman.py https://github.com/saltstack/salt/blob/develop/salt/modules/yumpkg.py https://github.com/saltstack/salt/blob/develop/salt/modules/apt.py

Documentation¶

Salt modules are self documenting, the sys.doc() function will return the documentation for all available modules:

salt '*' sys.doc

This function simply prints out the docstrings found in the modules; when writing Salt modules, please follow the formatting conventions for docstrings as they appear in the other modules.

Adding Documentation to Salt Modules¶

Since life is much better with documentation, it is strongly suggested that all Salt modules have documentation added. Any Salt modules submitted for inclusion in the main distribution of Salt will be required to have documentation.

Documenting Salt modules is easy! Just add a Python docstring to the function.

def spam(eggs):
    '''
    A function to make some spam with eggs!

    CLI Example::

        salt '*' test.spam eggs
    '''
    return eggs

Now when the sys.doc call is executed the docstring will be cleanly returned to the calling terminal.

Add Module metadata¶

Add information about the module using the following field lists:

:maintainer:    Thomas Hatch <thatch@saltstack.com, Seth House <shouse@saltstack.com>
:maturity:      new
:depends:       python-mysqldb
:platform:      all

The maintainer field is a comma-delimited list of developers who help maintain this module.

The maturity field indicates the level of quality and testing for this module. Standard labels will be determined.

The depends field is a comma-delimited list of modules that this module depends on.

The platform field is a comma-delimited list of platforms that this module is known to run on.

How Functions are Read¶

In Salt, Python callable objects contained within a module are made available to the Salt minion for use. The only exception to this rule is a callable object with a name starting with an underscore _.

Objects Loaded Into the Salt Minion¶

def foo(bar):
    return bar

class baz:
    def __init__(self, quo):
        pass

Objects NOT Loaded into the Salt Minion¶

def _foobar(baz): # Preceded with an _
    return baz

cheese = {} # Not a callable Python object

Useful Decorators for Modules¶

Sometimes when writing modules for large scale deployments you run into some small things that end up severely complicating the code. To alleviate some of this pain Salt has some useful decorators for use within modules!

Depends Decorator¶

When writing custom modules there are many times where some of the module will work on all hosts, but some functions require (for example) a service to be installed. Instead of trying to wrap much of the code in large try/except blocks you can use a simple decorator to do this. If the dependencies passed to the decorator don't exist, then the salt minion will remove those functions from the module on that host. If a "fallback_funcion" is defined, it will replace the function instead of removing it

from salt.utils.decorators import depends
try:
    import dependency_that_sometimes_exists
except ImportError:
    pass

@depends('dependency_that_sometimes_exists')
def foo():
    '''
    Function with a dependency on the "dependency_that_sometimes_exists" module,
    if the "dependency_that_sometimes_exists" is missing this function will not exist
    '''
    return True

def _fallback():
    '''
    Fallback function for the depends decorator to replace a function with
    '''
    return '"dependency_that_sometimes_exists" needs to be installed for this function to exist'

@depends('dependency_that_sometimes_exists', fallback_funcion=_fallback)
def foo():
    '''
    Function with a dependency on the "dependency_that_sometimes_exists" module.
    If the "dependency_that_sometimes_exists" is missing this function will be
    replaced with "_fallback"
    '''
    return True

Examples of Salt Modules¶

The existing Salt modules should be fairly easy to read and understand, the goal of the main distribution's Salt modules is not only to build a set of functions for Salt, but to stand as examples for building out more Salt modules.

The existing modules can be found here: https://github.com/saltstack/salt/blob/develop/salt/modules

The most simple module is the test module, it contains the simplest Salt function, test.ping:

def ping():
    '''
    Just used to make sure the minion is up and responding
    Return True

    CLI Example::

        salt '*' test.ping
    '''
    return True

Full list of builtin execution modules¶

Virtual modules

salt.modules.pkg¶

pkg is a virtual module that is fulfilled by one of the following modules:

salt.modules.sys¶

The regular salt modules execute in a separate context from the salt minion and manipulating the actual salt modules needs to happen in a higher level context within the minion process. This is where the sys pseudo module is used.

The sys pseudo module comes with a few functions that return data about the available functions on the minion or allows for the minion modules to be refreshed. These functions are as follows:

salt.modules.sys.doc([module[, module.function]])¶: Display the inline documentation for all available modules, or for the specified module or function.

salt.modules.sys.reload_modules()¶: Instruct the minion to reload all available modules in memory. This function can be called if the modules need to be re-evaluated for availability or new modules have been made available to the minion.

salt.modules.sys.list_modules()¶: List all available (loaded) modules.

salt.modules.sys.list_functions()¶: List all known functions that are in available (loaded) modules.

`aliases`	Manage the information in the aliases file
`alternatives`	Support for Alternatives system
`apache`	Support for Apache
`apt`	Support for APT (Advanced Packaging Tool)
`archive`	A module to wrap archive calls
`at`	Wrapper module for at(1)
`augeas_cfg`	Manages configuration files via augeas
`bluez`	Support for Bluetooth (using BlueZ in Linux).
`brew`	Homebrew for Mac OS X
`bridge`	Module for gathering and managing bridging information
`bsd_shadow`	Manage the password database on BSD systems
`cassandra`	Cassandra NoSQL Database Module
`cmdmod`	A module for shelling out
`config`	Return config information
`cp`	Minion side functions for salt-cp
`cron`	Work with cron
`daemontools`	daemontools service module. This module will create daemontools type
`darwin_sysctl`	Module for viewing and modifying sysctl parameters
`data`	Manage a local persistent data structure that can hold any arbitrary data
`ddns`	Support for RFC 2136 dynamic DNS updates.
`debconfmod`	Support for Debconf
`debian_service`	Service support for Debian systems (uses update-rc.d and /sbin/service)
`dig`	Compendium of generic DNS utilities
`disk`	Module for gathering disk information
`djangomod`	Manage Django sites
`dnsmasq`	Module for managing dnqmasq
`dnsutil`	Compendium of generic DNS utilities
`dpkg`	Support for DEB packages
`ebuild`	Support for Portage
`eix`	Support for Eix
`eselect`	Support for eselect, Gentoo's configuration and management tool.
`event`	Use the Salt Event System to fire events from the master to the minion and vice-versa.
`extfs`	Module for managing ext2/3/4 file systems
`file`	Manage information about regular files, directories,
`freebsd_sysctl`	Module for viewing and modifying sysctl parameters
`freebsdjail`	The jail module for FreeBSD
`freebsdkmod`	Module to manage FreeBSD kernel modules
`freebsdpkg`	Package support for FreeBSD
`freebsdservice`	The service module for FreeBSD
`gem`	Manage ruby gems.
`gentoo_service`	Top level package command wrapper, used to translate the os detected by grains to the correct service manager
`gentoolkitmod`	Support for Gentoolkit
`git`	Support for the Git SCM
`glance`	Module for handling openstack glance calls.
`grains`	Return/control aspects of the grains data
`groupadd`	Manage groups on Linux and OpenBSD
`grub_legacy`	Support for GRUB Legacy
`guestfs`	Interact with virtual machine images via libguestfs
`hg`	Support for the Mercurial SCM
`hosts`	Manage the information in the hosts file
`img`	Virtual machine image management tools
`iptables`	Support for iptables
`key`	Functions to view the minion's public key information
`keyboard`	Module for managing keyboards on POSIX-like systems.
`keystone`	Module for handling openstack keystone calls.
`kmod`	Module to manage Linux kernel modules
`launchctl`	Module for the management of MacOS systems that use launchd/launchctl
`layman`	Support for Layman
`ldapmod`	Salt interface to LDAP commands
`linux_acl`	Support for Linux File Access Control Lists
`linux_lvm`	Support for Linux LVM2
`linux_sysctl`	Module for viewing and modifying sysctl parameters
`localemod`	Module for managing locales on POSIX-like systems.
`locate`	Module for using the locate utilities
`logrotate`	Module for managing logrotate.
`lxc`	Work with linux containers
`makeconf`	Support for modifying make.conf under Gentoo
`match`	The match module allows for match routines to be run and determine target specs
`mdadm`	Salt module to manage RAID arrays with mdadm
`mine`	The function cache system allows for data to be stored on the master so it can be easily read by other minions
`modjk`	Control Modjk via the Apache Tomcat "Status" worker
`mongodb`	Module to provide MongoDB functionality to Salt
`monit`	Monit service module.
`moosefs`	Module for gathering and managing information about MooseFS
`mount`	Salt module to manage unix mounts and the fstab file
`munin`	Run munin plugins/checks from salt and format the output as data.
`mysql`	Module to provide MySQL compatibility to salt.
`netbsd_sysctl`	Module for viewing and modifying sysctl parameters
`netbsdservice`	The service module for NetBSD
`network`	Module for gathering and managing network information
`nfs3`	Module for managing NFS version 3.
`nginx`	Support for nginx
`nova`	Module for handling openstack nova calls.
`npm`	Manage and query NPM packages.
`nzbget`	Support for nzbget
`openbsdpkg`	Package support for OpenBSD
`openbsdservice`	The service module for OpenBSD
`osxdesktop`	Mac OS X implementations of various commands in the "desktop" interface
`pacman`	A module to wrap pacman calls, since Arch is the best
`pam`	Support for pam
`parted`	Module for managing partitions on POSIX-like systems.
`pecl`	Manage PHP pecl extensions.
`pillar`	Extract the pillar data for this minion
`pip`	Install Python packages with pip to either the system or a virtualenv
`pkg_resource`	Resources needed by pkg providers
`pkgin`	Package support for pkgin based systems, inspired from freebsdpkg module
`pkgng`	Support for pkgng
`pkgutil`	Pkgutil support for Solaris
`portage_config`	Configure `portage(5)`
`postgres`	Module to provide Postgres compatibility to salt.
`poudriere`	Support for poudriere
`ps`	A salt interface to psutil, a system and process library.
`publish`	Publish a command from a minion to a target
`puppet`	Execute puppet routines
`pw_group`	Manage groups on FreeBSD
`pw_user`	Manage users with the useradd command
`qemu_img`	Qemu-img Command Wrapper
`qemu_nbd`	Qemu Command Wrapper
`quota`	Module for managing quotas on POSIX-like systems.
`rabbitmq`	Module to provide RabbitMQ compatibility to Salt.
`rbenv`	Manage ruby installations with rbenv.
`reg`	Manage the registry on Windows
`ret`	Module to integrate with the returner system and retrieve data sent to a salt returner
`rh_ip`	The networking module for RHEL/Fedora based distros
`rh_service`	Service support for RHEL-based systems, including support for both upstart and sysvinit
`rpm`	Support for rpm
`rvm`	Manage ruby installations and gemsets with RVM, the Ruby Version Manager.
`s3`	Connection module for Amazon S3
`saltutil`	The Saltutil module is used to manage the state of the salt minion itself.
`seed`	Virtual machine image management tools
`selinux`	Execute calls on selinux
`service`	The default service module, if not otherwise specified salt will fall back
`shadow`	Manage the shadow file
`smartos_imgadm`	Module for running imgadm command on SmartOS
`smartos_vmadm`	Module for managing VMs on SmartOS
`smf`	Service support for Solaris 10 and 11, should work with other systems that use SMF also.
`solaris_group`	Manage groups on Solaris
`solaris_shadow`	Manage the password database on Solaris systems
`solaris_user`	Manage users with the useradd command
`solarispkg`	Package support for Solaris
`solr`	Apache Solr Salt Module
`sqlite3`	Support for SQLite3
`ssh`	Manage client ssh components
`state`	Control the state system on the minion
`status`	Module for returning various status data about a minion.
`supervisord`	Provide the service module for system supervisord or supervisord in a virtualenv
`svn`	Subversion SCM
`sysbench`	The 'sysbench' module is used to analyse the performance of the minions, right from the master! It measures various system parameters such as CPU, Memory, FileI/O, Threads and Mutex.
`sysmod`	The sys module provides information about the available functions on the minion
`system`	Support for reboot, shutdown, etc
`systemd`	Provide the service module for systemd
`test`	Module for running arbitrary tests
`timezone`	Module for managing timezone on POSIX-like systems.
`tls`	A salt module for SSL/TLS.
`tomcat`	Support for Tomcat
`upstart`	Module for the management of upstart systems.
`useradd`	Manage users with the useradd command
`virt`	Work with virtual machines managed by libvirt
`virtualenv_mod`	Create virtualenv environments
`win_disk`	Module for gathering disk information on Windows
`win_file`	Manage information about files on the minion, set/read user, group
`win_groupadd`	Manage groups on Windows
`win_network`	Module for gathering and managing network information
`win_pkg`	A module to manage software on Windows
`win_service`	Windows Service module.
`win_shadow`	Manage the shadow file
`win_status`	Module for returning various status data about a minion.
`win_system`	Support for reboot, shutdown, etc
`win_useradd`	Manage Windows users with the net user command
`xapi`	This module (mostly) uses the XenAPI to manage Xen virtual machines.
`yumpkg`	Support for YUM
`yumpkg5`	Support for YUM
`zfs`	Module for running ZFS command
`zpool`	Module for running ZFS zpool command
`zypper`	Package support for openSUSE via the zypper package manager

Returners¶

By default the return values of the commands sent to the Salt minions are returned to the salt-master. But since the commands executed on the Salt minions are detached from the call on the Salt master, anything at all can be done with the results data.

This is where the returner interface comes in. Returners are modules called in addition to returning the data to the Salt master.

The returner interface allows the return data to be sent to any system that can receive data. This means that return data can be sent to a Redis server, a MongoDB server, a MySQL server, or any system!

See also

Full list of builtin returners

Using Returners¶

All commands will return the command data back to the master. Adding more returners will ensure that the data is also sent to the specified returner interfaces.

Specifying what returners to use is done when the command is invoked:

salt '*' test.ping --return redis_return

This command will ensure that the redis_return returner is used.

It is also possible to specify multiple returners:

salt '*' test.ping --return mongo_return,redis_return,cassandra_return

In this scenario all three returners will be called and the data from the test.ping command will be sent out to the three named returners.

Writing a Returner¶

A returner is a module which contains a returner function, the returner function must accept a single argument. this argument is the return data from the called minion function. So if the minion function test.ping is called the value of the argument will be True.

A simple returner is implemented here:

import redis
import json

def returner(ret):
    '''
    Return information to a redis server
    '''
    # Get a redis connection
    serv = redis.Redis(
                host='redis-serv.example.com',
                port=6379,
                db='0')
    serv.sadd("%(id)s:jobs" % ret, ret['jid'])
    serv.set("%(jid)s:%(id)s" % ret, json.dumps(ret['return']))
    serv.sadd('jobs', ret['jid'])
    serv.sadd(ret['jid'], ret['id'])

This simple example of a returner set to send the data to a redis server serializes the data as json and sets it in redis.

You can place your custom returners in a _returners directory within the file_roots specified by the master config file. These custom returners are distributed when state.highstate is run, or by executing the saltutil.sync_returners or saltutil.sync_all functions.

Any custom returners which have been synced to a minion, that are named the same as one of Salt's default set of returners, will take the place of the default returner with the same name. Note that a returner's default name is its filename (i.e. foo.py becomes returner foo), but that its name can be overridden by using a __virtual__ function. A good example of this can be found in the redis returner, which is named redis_return.py but is loaded as simply redis:

try:
    import redis
    HAS_REDIS = True
except ImportError:
    HAS_REDIS = False

def __virtual__():
    if not HAS_REDIS:
        return False
    return 'redis'

Examples¶

The collection of built-in Salt returners can be found here: https://github.com/saltstack/salt/blob/develop/salt/returners

Full list of builtin returner modules¶

`carbon_return`	Take data from salt and "return" it into a carbon receiver
`cassandra_return`	Return data to a Cassandra ColumnFamily
`local`	The local returner is used to test the returner interface, it just prints the
`mongo_future_return`	Return data to a mongodb server
`mongo_return`	Return data to a mongodb server
`mysql`	Return data to a mysql server
`postgres`	Return data to a postgresql server
`redis_return`	Return data to a redis server
`sentry_return`	Salt returner that report execution results back to sentry.
`smtp_return`	Return salt data via email
`sqlite3_return`	Insert minion return data into a sqlite3 database
`syslog_return`	Return data to the host operating system's syslog facility

File State Backups¶

In 0.10.2 a new feature was added for backing up files that are replaced by the file.managed and file.recurse states. The new feature is called the backup mode. Setting the backup mode is easy, but is can be set in a number of places.

The backup_mode can be set in the minion config file:

backup_mode: minion

Or it can be set for each file:

/etc/ssh/sshd_config:
  file.managed:
    - source: salt://ssh/sshd_config
    - backup: minion

Backed-up Files¶

The files will be saved in the minion cachedir under the directory named file_backup. The files will be in the location relative to where they were under the root filesystem and be appended with a timestamp. This should make them easy to browse.

Interacting with Backups¶

Starting with version 0.17.0, it will be possible to list, restore, and delete previously-created backups.

Listing¶

The backups for a given file can be listed using file.list_backups:

# salt foo.bar.com file.list_backups /tmp/foo.txt
foo.bar.com:
    ----------
    0:
        ----------
        Backup Time:
            Sat Jul 27 2013 17:48:41.738027
        Location:
            /var/cache/salt/minion/file_backup/tmp/foo.txt_Sat_Jul_27_17:48:41_738027_2013
        Size:
            13
    1:
        ----------
        Backup Time:
            Sat Jul 27 2013 17:48:28.369804
        Location:
            /var/cache/salt/minion/file_backup/tmp/foo.txt_Sat_Jul_27_17:48:28_369804_2013
        Size:
            35

Restoring¶

Restoring is easy using file.restore_backup, just pass the path and the numeric id found with file.list_backups:

# salt foo.bar.com file.restore_backup /tmp/foo.txt 1
foo.bar.com:
    ----------
    comment:
        Successfully restored /var/cache/salt/minion/file_backup/tmp/foo.txt_Sat_Jul_27_17:48:28_369804_2013 to /tmp/foo.txt
    result:
        True

The existing file will be backed up, just in case, as can be seen if file.list_backups is run again:

# salt foo.bar.com file.list_backups /tmp/foo.txt
foo.bar.com:
    ----------
    0:
        ----------
        Backup Time:
            Sat Jul 27 2013 18:00:19.822550
        Location:
            /var/cache/salt/minion/file_backup/tmp/foo.txt_Sat_Jul_27_18:00:19_822550_2013
        Size:
            53
    1:
        ----------
        Backup Time:
            Sat Jul 27 2013 17:48:41.738027
        Location:
            /var/cache/salt/minion/file_backup/tmp/foo.txt_Sat_Jul_27_17:48:41_738027_2013
        Size:
            13
    2:
        ----------
        Backup Time:
            Sat Jul 27 2013 17:48:28.369804
        Location:
            /var/cache/salt/minion/file_backup/tmp/foo.txt_Sat_Jul_27_17:48:28_369804_2013
        Size:
            35

Note

Since no state is being run, restoring a file will not trigger any watches for the file. So, if you are restoring a config file for a service, it will likely still be necessary to run a service.restart.

Deleting¶

Deleting backups can be done using mod:file.delete_backup <salt.modules.file.delete_backup>:

# salt foo.bar.com file.delete_backup /tmp/foo.txt 0
foo.bar.com:
    ----------
    comment:
        Successfully removed /var/cache/salt/minion/file_backup/tmp/foo.txt_Sat_Jul_27_18:00:19_822550_2013
    result:
        True

Extending External SLS Data¶

Sometimes a state defined in one SLS file will need to be modified from a separate SLS file. A good example of this is when an argument needs to be overwritten or when a service needs to watch an additional state.

The Extend Declaration¶

The standard way to extend is via the extend declaration. The extend declaration is a top level declaration like include and encapsulates ID declaration data included from other SLS files. A standard extend looks like this:

include:
  - http
  - ssh

extend:
  apache:
    file:
      - name: /etc/httpd/conf/httpd.conf
      - source: salt://http/httpd2.conf
  ssh-server:
    service:
      - watch:
        - file: /etc/ssh/banner

/etc/ssh/banner:
  file.managed:
    - source: salt://ssh/banner

A few critical things happened here, first off the SLS files that are going to be extended are included, then the extend dec is defined. Under the extend dec 2 IDs are extended, the apache ID's file state is overwritten with a new name and source. Than the ssh server is extended to watch the banner file in addition to anything it is already watching.

Extend is a Top Level Declaration¶

This means that extend can only be called once in an sls, if if is used twice then only one of the extend blocks will be read. So this is WRONG:

include:
  - http
  - ssh

extend:
  apache:
    file:
      - name: /etc/httpd/conf/httpd.conf
      - source: salt://http/httpd2.conf
# Second extend will overwrite the first!! Only make one
extend:
  ssh-server:
    service:
      - watch:
        - file: /etc/ssh/banner

The Requisite "in" Statement¶

Since one of the most common things to do when extending another SLS is to add states for a service to watch, or anything for a watcher to watch, the requisite in statement was added to 0.9.8 to make extending the watch and require lists easier. The ssh-server extend statement above could be more cleanly defined like so:

include:
  - ssh

/etc/ssh/banner:
  file.managed:
    - source: salt://ssh/banner
    - watch_in:
      - service: ssh-server

Rules to Extend By¶

There are a few rules to remember when extending states:

Always include the SLS being extended with an include declaration
Requisites (watch and require) are appended to, everything else is overwritten
extend is a top level declaration, like an ID declaration, cannot be declared twice in a single SLS
Many IDs can be extended under the extend declaration

Failhard Global Option¶

Normally, when a state fails Salt continues to execute the remainder of the defined states and will only refuse to execute states that require the failed state.

But the situation may exist, where you would want all state execution to stop if a single state execution fails. The capability to do this is called failing hard.

State Level Failhard¶

A single state can have a failhard set, this means that if this individual state fails that all state execution will immediately stop. This is a great thing to do if there is a state that sets up a critical config file and setting a require for each state that reads the config would be cumbersome. A good example of this would be setting up a package manager early on:

/etc/yum.repos.d/company.repo:
  file.managed:
    - source: salt://company/yumrepo.conf
    - user: root
    - group: root
    - mode: 644
    - order: 1
    - failhard: True

In this situation, the yum repo is going to be configured before other states, and if it fails to lay down the config file, than no other states will be executed.

Global Failhard¶

It may be desired to have failhard be applied to every state that is executed, if this is the case, then failhard can be set in the master configuration file. Setting failhard in the master configuration file will result in failing hard when any minion gathering states from the master have a state fail.

This is NOT the default behavior, normally Salt will only fail states that require a failed state.

Using the global failhard is generally not recommended, since it can result in states not being executed or even checked. It can also be confusing to see states failhard if an admin is not actively aware that the failhard has been set.

To use the global failhard set failhard: True in the master configuration file.

Highstate data structure definitions¶

The Salt State Tree¶

Top file

The main state file that instructs minions what environment and modules to use during state execution.

Configurable via state_top.

See also

A detailed description of the top file

State tree: A collection of SLS files that live under the directory specified in file_roots. A state tree can be organized into SLS modules.

Include declaration¶

Include declaration

Defines a list of module reference strings to include in this SLS.

Occurs only in the top level of the highstate structure.

Example:

include:
  - edit.vim
  - http.server

Module reference¶

Module reference: The name of a SLS module defined by a separate SLS file and residing on the Salt Master. A module named edit.vim is a reference to the SLS file salt://edit/vim.sls.

ID declaration¶

ID declaration

Defines an individual highstate component. Always references a value of a dictionary containing keys referencing state declarations and requisite declarations. Can be overridden by a name declaration or a names declaration.

Occurs on the top level or under the extend declaration.

Must be unique across entire state tree. If the same ID declaration is used twice, only the first one matched will be used. All subsequent ID declarations with the same name will be ignored.

Note

Naming gotchas

Until 0.9.6, IDs could not contain a dot, otherwise highstate summary output was unpredictable. (It was fixed in versions 0.9.7 and above)

Extend declaration¶

Extend declaration

Extends a name declaration from an included SLS module. The keys of the extend declaration always define existing ID declarations which have been defined in included SLS modules.

Occurs only in the top level and defines a dictionary.

Extend declarations are useful for adding-to or overriding parts of a state declaration that is defined in another SLS file. In the following contrived example, the shown mywebsite.sls file is include -ing and extend -ing the apache.sls module in order to add a watch declaration that will restart Apache whenever the Apache configuration file, mywebsite changes.

include:
  - apache

extend:
  apache:
    service:
      - watch:
        - file: mywebsite

mywebsite:
  file:
    - managed

See also

watch_in and require_in

Sometimes it is more convenient to use the watch_in or require_in syntax instead of extending another SLS file.

State Requisites

State declaration¶

State declaration

A list which contains one string defining the function declaration and any number of function arg declaration dictionaries.

Can, optionally, contain a number of additional components like the name override components — name and names. Can also contain requisite declarations.

Occurs under an ID declaration.

Requisite declaration¶

Requisite declaration

A list containing requisite references.

Used to build the action dependency tree. While Salt states are made to execute in a deterministic order, this order is managed by requiring and watching other Salt states.

Occurs as a list component under a state declaration or as a key under an ID declaration.

Requisite reference¶

Requisite reference

A single key dictionary. The key is the name of the referenced state declaration and the value is the ID of the referenced ID declaration.

Occurs as a single index in a requisite declaration list.

Function declaration¶

Function declaration

The name of the function to call within the state. A state declaration can contain only a single function declaration.

For example, the following state declaration calls the installed function in the pkg state module:

httpd:
  pkg.installed

The function can be declared inline with the state as a shortcut, but the actual data structure is better referenced in this form:

httpd:
  pkg:
    - installed

Where the function is a string in the body of the state declaration. Technically when the function is declared in dot notation the compiler converts it to be a string in the state declaration list. Note that the use of the first example more than once in an ID declaration is invalid yaml.

INVALID:

httpd:
  pkg.installed
  service.running

When passing a function without arguments and another state declaration within a single ID declaration, then the long or "standard" format needs to be used since otherwise it does not represent a valid data structure.

VALID:

httpd:
  pkg:
    - installed
  service:
    - running

Occurs as the only index in the state declaration list.

Function arg declaration¶

Function arg declaration

A single key dictionary referencing a Python type which is to be passed to the named function declaration as a parameter. The type must be the data type expected by the function.

Occurs under a function declaration.

For example in the following state declaration user, group, and mode are passed as arguments to the managed function in the file state module:

/etc/http/conf/http.conf:
  file.managed:
    - user: root
    - group: root
    - mode: 644

Name declaration¶

Name declaration

Overrides the name argument of a state declaration. If name is not specified the ID declaration satisfies the name argument.

The name is always a single key dictionary referencing a string.

Overriding name is useful for a variety of scenarios.

For example, avoiding clashing ID declarations. The following two state declarations cannot both have /etc/motd as the ID declaration:

motd_perms:
  file.managed:
    - name: /etc/motd
    - mode: 644

motd_quote:
  file.append:
    - name: /etc/motd
    - text: "Of all smells, bread; of all tastes, salt."

Another common reason to override name is if the ID declaration is long and needs to be referenced in multiple places. In the example below it is much easier to specify mywebsite than to specify /etc/apache2/sites-available/mywebsite.com multiple times:

mywebsite:
  file.managed:
    - name: /etc/apache2/sites-available/mywebsite.com
    - source: salt://mywebsite.com

a2ensite mywebsite.com:
  cmd.wait:
    - unless: test -L /etc/apache2/sites-enabled/mywebsite.com
    - watch:
      - file: mywebsite

apache2:
  service:
    - running
    - watch:
      - file: mywebsite

Names declaration¶

Names declaration: Expands the contents of the containing state declaration into multiple state declarations, each with its own name.

For example, given the following state declaration:

python-pkgs:
  pkg.installed:
    - names:
      - python-django
      - python-crypto
      - python-yaml

Once converted into the lowstate data structure the above state declaration will be expanded into the following three state declarations:

python-django:
  pkg.installed

python-crypto:
  pkg.installed

python-yaml:
  pkg.installed

Large example¶

Here is the layout in yaml using the names of the highdata structure components.

<Include Declaration>:
  - <Module Reference>
  - <Module Reference>

<Extend Declaration>:
  <ID Declaration>:
    [<overrides>]


# standard declaration

<ID Declaration>:
  <State Declaration>:
    - <Function>
    - <Function Arg>
    - <Function Arg>
    - <Function Arg>
    - <Name>: <name>
    - <Requisite Declaration>:
      - <Requisite Reference>
      - <Requisite Reference>


# inline function and names

<ID Declaration>:
  <State Declaration>.<Function>:
    - <Function Arg>
    - <Function Arg>
    - <Function Arg>
    - <Names>:
      - <name>
      - <name>
      - <name>
    - <Requisite Declaration>:
      - <Requisite Reference>
      - <Requisite Reference>


# multiple states for single id

<ID Declaration>:
  <State Declaration>:
    - <Function>
    - <Function Arg>
    - <Name>: <name>
    - <Requisite Declaration>:
      - <Requisite Reference>
  <State Declaration>:
    - <Function>
    - <Function Arg>
    - <Names>:
      - <name>
      - <name>
    - <Requisite Declaration>:
      - <Requisite Reference>

Include and Exclude¶

Salt sls files can include other sls files and exclude sls files that have been otherwise included. This allows for an sls file to easily extend or manipulate other sls files.

Include¶

When other sls files are included, everything defined in the included sls file will be added to the state run. When including define a list of sls formulas to include:

include:
  - http
  - libvirt

The include statement will include sls formulas from the same environment that the including sls formula is in. But the environment can be explicitly defined in the configuration to override the running environment, therefore if an sls formula needs to be included from an external environment named "dev" the following syntax is used:

include:
  - dev: http

Relative Include¶

In Salt 0.16.0 the capability to include sls formulas which are relative to the running sls formula was added, simply precede the formula name with a .:

include:
  - .virt
  - .virt.hyper

Exclude¶

The exclude statement, added in Salt 0.10.3 allows an sls to hard exclude another sls file or a specific id. The component is excluded after the high data has been compiled, so nothing should be able to override an exclude.

Since the exclude can remove an id or an sls the type of component to exclude needs to be defined. an exclude statement that verifies that the running highstate does not contain the http sls and the /etc/vimrc id would look like this:

exclude:
  - sls: http
  - id: /etc/vimrc

State Enforcement¶

Salt offers an optional interface to manage the configuration or "state" of the Salt minions. This interface is a fully capable mechanism used to enforce the state of systems from a central manager.

The Salt state system is made to be accurate, simple, and fast. And like the rest of the Salt system, Salt states are highly modular.

State management¶

State management, also frequently called software configuration management (SCM), is a program that puts and keeps a system into a predetermined state. It installs software packages, starts or restarts services, or puts configuration files in place and watches them for changes.

Having a state management system in place allows you to easily and reliably configure and manage a few servers or a few thousand servers. It allows you to keep that configuration under version control.

Salt States is an extension of the Salt Modules that we discussed in the previous remote execution tutorial. Instead of calling one-off executions the state of a system can be easily defined and then enforced.

Understanding the Salt State System Components¶

The Salt state system is comprised of a number of components. As a user, an understanding of the SLS and renderer systems are needed. But as a developer, an understanding of Salt states and how to write the states is needed as well.

Salt SLS System¶

SLS

The primary system used by the Salt state system is the SLS system. SLS stands for SaLt State.

The Salt States are files which contain the information about how to configure Salt minions. The states are laid out in a directory tree and can be written in many different formats.

The contents of the files and they way they are laid out is intended to be as simple as possible while allowing for maximum flexibility. The files are laid out in states and contains information about how the minion needs to be configured.

SLS File Layout¶

SLS files are laid out in the Salt file server. A simple layout can look like this:

top.sls
ssh.sls
sshd_config
users/init.sls
users/admin.sls
salt/init.sls
salt/master.sls

This example shows the core concepts of file layout. The top file is a key component and is used with Salt matchers to match SLS states with minions. The .sls files are states. The rest of the files are seen by the Salt master as just files that can be downloaded.

The states are translated into dot notation, so the ssh.sls file is seen as the ssh state, the users/admin.sls file is seen as the users.admin states.

The init.sls files are translated to be the state name of the parent directory, so the salt/init.sls file translates to the Salt state.

The plain files are visible to the minions, as well as the state files. In Salt, everything is a file; there is no "magic translation" of files and file types. This means that a state file can be distributed to minions just like a plain text or binary file.

SLS Files¶

The Salt state files are simple sets of data. Since the SLS files are just data they can be represented in a number of different ways. The default format is yaml generated from a Jinja template. This allows for the states files to have all the language constructs of Python and the simplicity of yaml. State files can then be complicated Jinja templates that translate down to yaml, or just plain and simple yaml files!

The State files are constructed data structures in a simple format. The format allows for many real activates to be expressed in very little text, while maintaining the utmost in readability and usability.

Here is an example of a Salt State:

vim:
  pkg:
    - installed

salt:
  pkg:
    - latest
  service.running:
    - require:
      - file: /etc/salt/minion
      - pkg: salt
    - names:
      - salt-master
      - salt-minion
    - watch:
      - file: /etc/salt/minion

/etc/salt/minion:
  file.managed:
    - source: salt://salt/minion
    - user: root
    - group: root
    - mode: 644
    - require:
      - pkg: salt

This short stanza will ensure that vim is installed, Salt is installed and up to date, the salt-master and salt-minion daemons are running and the Salt minion configuration file is in place. It will also ensure everything is deployed in the right order and that the Salt services are restarted when the watched file updated.

The Top File¶

The top file is the mapping for the state system. The top file specifies which minions should have which modules applied and which environments they should draw the states from.

The top file works by specifying the environment, containing matchers with lists of Salt states sent to the matching minions:

base:
  '*':
    - salt
    - users
    - users.admin
  'saltmaster.*':
    - match: pcre
    - salt.master

This simple example uses the base environment, which is built into the default Salt setup, and then all minions will have the modules salt, users and users.admin since '*' will match all minions. Then the regular expression matcher will match all minions' with an id matching saltmaster.* and add the salt.master state.

Renderer System¶

The Renderer system is a key component to the state system. SLS files are representations of Salt "high data" structures. All Salt cares about when reading an SLS file is the data structure that is produced from the file.

This allows Salt states to be represented by multiple types of files. The Renderer system can be used to allow different formats to be used for SLS files.

The available renderers can be found in the renderers directory in the Salt source code:

https://github.com/saltstack/salt/blob/develop/salt/renderers

By default SLS files are rendered using Jinja as a templating engine, and yaml as the serialization format. Since the rendering system can be extended simply by adding a new renderer to the renderers directory, it is possible that any structured file could be used to represent the SLS files.

In the future XML will be added, as well as many other formats.

Reloading Modules¶

Some salt states require specific packages to be installed in order for the module to load, as an example the pip state module requires the pip package for proper name and version parsing. On most of the common cases, salt is clever enough to transparently reload the modules, for example, if you install a package, salt reloads modules because some other module or state might require just that package which was installed. On some edge-cases salt might need to be told to reload the modules. Consider the following state file which we'll call pep8.sls:

python-pip:
  cmd:
    - run
    - cwd: /
    - name: easy_install --script-dir=/usr/bin -U pip

pep8:
  pip.installed
  requires:
    - cmd: python-pip

The above example installs pip using easy_install from setuptools and installs pep8 using pip, which, as told earlier, requires pip to be installed system-wide. Let's execute this state:

salt-call state.sls pep8

The execution output would be something like:

----------
    State: - pip
    Name:      pep8
    Function:  installed
        Result:    False
        Comment:   State pip.installed found in sls pep8 is unavailable

        Changes:

Summary
------------
Succeeded: 1
Failed:    1
------------
Total:     2

If we executed the state again the output would be:

----------
    State: - pip
    Name:      pep8
    Function:  installed
        Result:    True
        Comment:   Package was successfully installed
        Changes:   pep8==1.4.6: Installed

Summary
------------
Succeeded: 2
Failed:    0
------------
Total:     2

Since we installed pip using cmd, salt has no way to know that a system-wide package was installed. On the second execution, since the required pip package was installed, the state executed perfectly.

To those thinking, couldn't salt reload modules on every state step since it already does for some cases? It could, but it should not since it would greatly slow down state execution.

So how do we solve this edge-case? reload_modules!

reload_modules is a boolean option recognized by salt on all available states which, does exactly what it tells use, forces salt to reload it's modules once that specific state finishes. The fixed state file would now be:

python-pip:
  cmd:
    - run
    - cwd: /
    - name: easy_install --script-dir=/usr/bin -U pip
    - reload_modules: true

pep8:
  pip.installed
  requires:
    - cmd: python-pip

Let's run it, once:

salt-call state.sls pep8

And it's output now is:

----------
    State: - pip
    Name:      pep8
    Function:  installed
        Result:    True
        Comment:   Package was successfully installed
        Changes:   pep8==1.4.6: Installed

Summary
------------
Succeeded: 2
Failed:    0
------------
Total:     2

State System Layers¶

The Salt state system is comprised of multiple layers. While using Salt does not require an understanding of the state layers, a deeper understanding of how Salt compiles and manages states can be very beneficial.

Function Call¶

The lowest layer of functionality in the state system is the direct state function call. State executions are executions of single state functions at the core. These individual functions are defined in state modules and can be called directly via the state.single command.

salt '*' state.single pkg.installed name='vim'

Low Chunk¶

The low chunk is the bottom of the Salt state compiler. This is a data representation of a single function call. The low chunk is sent to the state caller and used to execute a single state function.

A single low chunk can be executed manually via the state.low command.

salt '*' state.low '{name: vim, state: pkg, fun: installed}'

The passed data reflects what the state execution system gets after compiling the data down from sls formulas.

Low State¶

The Low State layer is the list of low chunks "evaluated" in order. To see what the low state looks like for a highstate, run:

salt '*' state.show_lowstate

This will display the raw lowstate in the order which each low chunk will be evaluated. The order of evaluation is not necessarily the order of execution, since requisites are evaluated at runtime. Requisite execution and evaluation is finite; this means that the order of execution can be ascertained with 100% certainty based on the order of the low state.

High Data¶

High data is the data structure represented in YAML via SLS files. The High data structure is created by merging the data components rendered inside sls files (or other render systems). The High data can be easily viewed by executing the state.show_highstate or state.show_sls functions. Since this data is a somewhat complex data structure, it may be easier to read using the json, yaml, or pprint outputters:

salt '*' state.show_highstate --out yaml
salt '*' state.show_sls edit.vim --out pprint

SLS¶

Above "High Data", the logical layers are no longer technically required to be executed, or to be executed in a hierarchy. This means that how the High data is generated is optional and very flexible. The SLS layer allows for many mechanisms to be used to render sls data from files or to use the fileserver backend to generate sls and file data from external systems.

The SLS layer can be called directly to execute individual sls formulas.

Note

SLS Formulas have historically been called "SLS files". This is because a single SLS was only constituted in a single file. Now the term "SLS Formula" better expresses how a compartmentalized SLS can be expressed in a much more dynamic way by combining pillar and other sources, and the SLS can be dynamically generated.

To call a single SLS formula named edit.vim, execute state.sls:

salt '*' state.sls edit.vim

HighState¶

Calling SLS directly logically assigns what states should be executed from the context of the calling minion. The Highstate layer is used to allow for full contextual assignment of what is executed where to be tied to groups of, or individual, minions entirely from the master. This means that the environment of a minion, and all associated execution data pertinent to said minion, can be assigned from the master without needing to execute or configure anything on the target minion. This also means that the minion can independently retrieve information about its complete configuration from the master.

To execute the High State call state.highstate:

salt '*' state.highstate

OverState¶

The overstate layer expresses the highest functional layer of Salt's automated logic systems. The Overstate allows for stateful and functional orchestration of routines from the master. The overstate defines in data execution stages which minions should execute states, or functions, and in what order using requisite logic.

Remote Control States¶

New in version 0.17.0.

Remote Control States is the capability to organize routines on minions from the master, using state files.

This allows for the use of the Salt state system to execute state runs and function runs in a way more powerful than the overstate, will full command of the requisite and ordering systems inside of states.

Note

Remote Control States was added in 0.17.0 with the intent to eventually deprecate the overstate system in favor of this new, substantially more powerful system.

The Overstate will still be maintained for the forseable future.

Creating States Trigger Remote Executions¶

The new salt state module allows for these new states to be defined in such a way to call out to the salt and/or the salt-ssh remote execution systems, this also supports the addition of states to connect to remote embedded devices.

To create a state that calls out to minions simple specify the salt.state or salt.function states:

webserver_setup:
  salt.state:
    - tgt: 'web*'
    - highstate: True

This sls file can now be referenced by the state.sls runner the same way an sls is normally referenced, assuming the default configurtion with /srv/salt as the root of the state tree and the above file being saved as /srv/salt/webserver.sls, the state can be run from the master with the salt-run command:

salt-run state.sls webserver

This will execute the defined state to fire up the webserver routine.

Calling Multiple State Runs¶

All of the concepts of states exist so building something more complex is easy:

Note

As of Salt 0.17.0 states are run in the order in which they are defined, so the cmd.run defined below will always execute first

cmd.run:
  salt.function:
    - roster: scan
    - tgt: 10.0.0.0/24
    - arg:
      - 'bootstrap'

storage_setup:
  salt.state:
    - tgt: 'role:storage'
    - tgt_type: grain
    - sls: ceph

webserver_setup:
  salt.state:
    - tgt: 'web*'
    - highstate: True

Ordering States¶

The way in which configuration management systems are executed is a hotly debated topic in the configuration management world. Two major philosophies exist on the subject, to either execute in an imperative fashion where things are executed in the order in which they are defined, or in a declarative fashion where dependencies need to be mapped between objects.

Imperative ordering is finite and generally considered easier to write, but declarative ordering is much more powerful and flexible but generally considered more difficult to create.

Salt has been created to get the best of both worlds. States are evaluated in a finite order, which guarantees that states are always executed in the same order, and the states runtime is declarative, making Salt fully aware of dependencies via the requisite system.

Also, in Salt 0.17.0, the state_auto_order option was added to Salt. It makes states get evaluated in the order in which they are defined.

State Auto Ordering¶

Salt always executes states in a finite manner, meaning that they will always execute in the same order regardless of the system that is executing them. But in Salt 0.17.0, the state_auto_order option was added. This option makes states get evaluated in the order in which they are defined in sls files.

The evaluation order makes it easy to know what order the states will be executed in, but it is important to note that the requisite system will override the ordering defined in the files, and the order option described below will also override the order in which states are defined in sls files.

If the classic ordering is preferred (lexicographic), then set state_auto_order to False in the master configuration file.

Requisite Statements¶

Note

This document represents behavior exhibited by Salt requisites as of version 0.9.7 of Salt.

Often when setting up states any single action will require or depend on another action. Salt allows you to build relationships between states with requisite statements. A requisite statement ensure that the named state is evaluated before the state requiring it. There are two types of requisite statements in Salt, require and watch.

These requisite statements are applied to a specific state declaration:

httpd:
  pkg:
    - installed
  file.managed:
    - name: /etc/httpd/conf/httpd.conf
    - source: salt://httpd/httpd.conf
    - require:
      - pkg: httpd

In this example we use the require requisite to declare that the file /etc/httpd/conf/httpd.conf should only be set up if the pkg state executes successfully.

The requisite system works by finding the states that are required and executing them before the state that requires them. Then the required states can be evaluated to see if they have executed correctly.

Note

Requisite matching

Requisites match on both the ID Declaration and the name parameter. Therefore, if you are using the pkgs or sources argument to install a list of packages in a pkg state, it's important to note that you cannot have a requisite that matches on an individual package in the list.

Multiple Requisites¶

The requisite statement is passed as a list, allowing for the easy addition of more requisites. Both requisite types can also be separately declared:

httpd:
  pkg:
    - installed
  service.running:
    - enable: True
    - watch:
      - file: /etc/httpd/conf/httpd.conf
    - require:
      - pkg: httpd
      - user: httpd
      - group: httpd
  file.managed:
    - name: /etc/httpd/conf/httpd.conf
    - source: salt://httpd/httpd.conf
    - require:
      - pkg: httpd
  user:
    - present
  group:
    - present

In this example the httpd service is only going to be started if the package, user, group and file are executed successfully.

The Require Requisite¶

The foundation of the requisite system is the require requisite. The require requisite ensures that the required state(s) are executed before the requiring state. So, if a state is declared that sets down a vimrc, then it would be pertinent to make sure that the vimrc file would only be set down if the vim package has been installed:

vim:
  pkg:
    - installed
  file.managed:
    - source: salt://vim/vimrc
    - require:
      - pkg: vim

In this case, the vimrc file will only be applied by Salt if and after the vim package is installed.

The Watch Requisite¶

The watch requisite is more advanced than the require requisite. The watch requisite executes the same logic as require (therefore if something is watched it does not need to also be required) with the addition of executing logic if the required states have changed in some way.

The watch requisite checks to see if the watched states have returned any changes. If the watched state returns changes, and the watched states execute successfully, then the watching state will execute a function that reacts to the changes in the watched states.

Perhaps an example can better explain the behavior:

redis:
  pkg:
    - latest
  file.managed:
    - source: salt://redis/redis.conf
    - name: /etc/redis.conf
    - require:
      - pkg: redis
  service.running:
    - enable: True
    - watch:
      - file: /etc/redis.conf
      - pkg: redis

In this example the redis service will only be started if the file /etc/redis.conf is applied, and the file is only applied if the package is installed. This is normal require behavior, but if the watched file changes, or the watched package is installed or upgraded, then the redis service is restarted.

Watch and the mod_watch Function¶

The watch requisite is based on the mod_watch function. Python state modules can include a function called mod_watch which is then called if the watch call is invoked. When mod_watch is called depends on the execution of the watched state, which:

If no changes then just run the watching state itself as usual. mod_watch is not called. This behavior is same as using a require.

If changes then run the watching state AND if that changes nothing then react by calling mod_watch.

When reacting, in the case of the service module the underlying service is restarted. In the case of the cmd state the command is executed.

The mod_watch function for the service state looks like this:

def mod_watch(name, sig=None, reload=False, full_restart=False):
    '''
    The service watcher, called to invoke the watch command.

    name
        The name of the init or rc script used to manage the service

    sig
        The string to search for when looking for the service process with ps
    '''
    if __salt__['service.status'](name, sig):
        if 'service.reload' in __salt__ and reload:
            restart_func = __salt__['service.reload']
        elif 'service.full_restart' in __salt__ and full_restart:
            restart_func = __salt__['service.full_restart']
        else:
            restart_func = __salt__['service.restart']
    else:
        restart_func = __salt__['service.start']

    result = restart_func(name)
    return {'name': name,
            'changes': {name: result},
            'result': result,
            'comment': 'Service restarted' if result else \
                       'Failed to restart the service'
           }

The watch requisite only works if the state that is watching has a mod_watch function written. If watch is set on a state that does not have a mod_watch function (like pkg), then the listed states will behave only as if they were under a require statement.

Also notice that a mod_watch may accept additional keyword arguments, which, in the sls file, will be taken from the same set of arguments specified for the state that includes the watch requisite. This means, for the earlier service.running example above, you can tell the service to reload instead of restart like this:

redis:

  # ... other state declarations omitted ...

    service.running:
      - enable: True
      - reload: True
      - watch:
        - file: /etc/redis.conf
        - pkg: redis

The Order Option¶

Before using the order option, remember that the majority of state ordering should be done with a requisite declaration, and that a requisite declaration will override an order option.

The order option is used by adding an order number to a state declaration with the option order:

vim:
  pkg.installed:
    - order: 1

By adding the order option to 1 this ensures that the vim package will be installed in tandem with any other state declaration set to the order 1.

Any state declared without an order option will be executed after all states with order options are executed.

But this construct can only handle ordering states from the beginning. Sometimes you may want to send a state to the end of the line. To do this, set the order to last:

vim:
  pkg.installed:
    - order: last

Remember that requisite statements override the order option. So the order option should be applied to the highest component of the requisite chain:

vim:
  pkg.installed:
    - order: last
    - require:
      - file: /etc/vimrc

/etc/vimrc:
  file.managed:
    - source: salt://edit/vimrc

OverState System¶

Often servers need to be set up and configured in a specific order, and systems should only be set up if systems earlier in the sequence has been set up without any issues.

The 0.11.0 release of Salt addresses this problem with a new layer in the state system called the Over State. The concept of the Over State is managed on the master, a series of state executions is controlled from the master and executed in order. If an execution requires that another execution first run without problems then the state executions will stop.

The Over State system is used to orchestrate deployment in a smooth and reliable way across multiple systems in small to large environments.

The Over State SLS¶

The overstate system is managed by an sls file located in the root of an environment. This file uses a data structure like all sls files.

The overstate sls file configures an unordered list of stages, each stage defines the minions to execute on and can define what sls files to run or to execute a state.highstate.

mysql:
  match: 'db*'
  sls:
    - mysql.server
    - drbd
webservers:
  match: 'web*'
  require:
    - mysql
all:
  match: '*'
  require:
    - mysql
    - webservers

The above defined over state will execute the mysql stage first because it is required by the webservers stage. The webservers stage will then be executed only if the mysql stage executes without any issues. The webservers stage will execute state.highstate on the matched minions, while the mysql stage will execute state.sls with the named sls files.

Finally the all stage will execute state.highstate on all systems only if the mysql and webservers stages complete without failures. The overstate system checks for any states that return a result of False, if the run has any False returns then the overstate will quit.

Adding Functions To Overstate¶

In 0.15.0 the ability to execute module functions directly in the overstate was added. Functions are called as a stage with the function key:

http:
  function:
    pkg.install:
      - http

The list of function arguments are passed after the declared function. Requisites only functions properly if the given function supports returning a custom return code.

Executing the Over State¶

The over state can be executed from the salt-run command, calling the state.over runner function. The function will by default look in the base environment for the overstate.sls file:

salt-run state.over

To specify the location of the overstate file and the environment to pull from pass the arguments to the salt-run command:

salt-run state.over base /root/overstate.sls

Remember, that these calls are made on the master.

State Providers¶

New in version 0.9.8.

Salt predetermines what modules should be mapped to what uses based on the properties of a system. These determinations are generally made for modules that provide things like package and service management.

Sometimes in states, it may be necessary to use an alternative module to provide the needed functionality. For instance, an older Arch Linux system may not be running systemd, so instead of using the systemd service module, you can revert to the default service module:

httpd:
  service.running:
    - enable: True
    - provider: service

In this instance, the basic service module (which manages sysvinit-based services) will replace the systemd module which is used by default on Arch Linux.

However, if it is necessary to make this override for most or every service, it is better to just override the provider in the minion config file, as described in the section below.

Setting a Provider in the Minion Config File¶

Sometimes, when running Salt on custom Linux spins, or distros that are derived from other distros, Salt does not successfully detect providers. The providers which are most likely to be affected by this are:

pkg
service
user
group

When something like this happens, rather than specifying the provider manually in each state, it easier to use the providers parameter in the minion config file to set the provider.

If you end up needing to override a provider because it was not detected, please let us know! File an issue on the issue tracker, and provide the output from the grains.items function, taking care to sanitize any sensitive information.

Below are tables that should help with deciding which provider to use if one needs to be overridden.

Provider: `pkg`¶

Execution Module	Used for
apt	Debian/Ubuntu-based distros which use `apt-get(8)` for package management
brew	Mac OS software management using Homebrew
ebuild	Gentoo-based systems (utilizes the `portage` python module as well as `emerge(1)`)
freebsdpkg	FreeBSD-based OSes using `pkg_add(1)`
openbsdpkg	OpenBSD-based OSes using `pkg_add(1)`
pacman	Arch Linux-based distros using `pacman(8)`
pkgin	NetBSD-based OSes using `pkgin(1)`
pkgng	FreeBSD-based OSes using `pkg(8)`
pkgutil	Solaris-based OSes using OpenCSW's `pkgutil(1)`
solarispkg	Solaris-based OSes using `pkgadd(1M)`
win_pkg	Windows
yumpkg	RedHat-based distros and derivatives (utilizes the `yum` and `rpmUtils` modules)
yumpkg5	RedHat-based distros and derivatives (wraps `yum(8)`)
zypper	SUSE-based distros using `zypper(8)`

Provider: `service`¶

Execution Module	Used for
debian_service	Debian Linux (non-systemd)
freebsdservice	FreeBSD-based OSes using `service(8)`
gentoo_service	Gentoo Linux using sysvinit and `rc-update(8)`
launchctl	Mac OS hosts using `launchctl(1)`
netbsdservice	NetBSD-based OSes
openbsdservice	OpenBSD-based OSes
rh_service	RedHat-based distros and derivatives using `service(8)` and `chkconfig(8)`. Supports both pure sysvinit and mixed sysvinit/upstart systems.
service	Fallback which simply wraps sysvinit scripts
smf	Solaris-based OSes which use SMF
systemd	Linux distros which use systemd
upstart	Ubuntu-based distros using upstart
win_service	Windows

Provider: `user`¶

Execution Module	Used for
useradd	Linux, NetBSD, and OpenBSD systems using `useradd(8)`, `userdel(8)`, and `usermod(8)`
pw_user	FreeBSD-based OSes using `pw(8)`
solaris_user	Solaris-based OSes using `useradd(1M)`, `userdel(1M)`, and `usermod(1M)`
win_useradd	Windows

Provider: `group`¶

Execution Module	Used for
groupadd	Linux, NetBSD, and OpenBSD systems using `groupadd(8)`, `groupdel(8)`, and `groupmod(8)`
pw_group	FreeBSD-based OSes using `pw(8)`
solaris_user	Solaris-based OSes using `groupadd(1M)`
win_groupadd	Windows

Arbitrary Module Redirects¶

The provider statement can also be used for more powerful means, instead of overwriting or extending the module used for the named service an arbitrary module can be used to provide certain functionality.

emacs:
  pkg.installed:
    - provider:
      - pkg: yumpkg5
      - cmd: customcmd

In this example the default pkg module is being redirected to use the yumpkg5 module (yum via shelling out instead of via the yum Python API), but is also using a custom module to invoke commands. This could be used to dramatically change the behavior of a given state.

Requisites¶

The Salt requisite system is used to create relationships between states. The core idea being that, when one state is dependent somehow on another, that inter-dependency can be easily defined.

Requisites come in two types. Direct requisites, and requisite_ins. The relationships are directional, so a requisite statement makes the requiring state declaration depend on the required state declaration:

vim:
  pkg.installed

/etc/vimrc:
  file.managed:
    - source: salt://edit/vimrc
    - require:
      - pkg: vim

So in this example, the file /etc/vimrc depends on the vim package.

Requisite_in statements are the opposite, instead of saying "I depend on something", requisite_ins say "Someone depends on me":

vim:
  pkg.installed:
    - require_in:
      - file: /etc/vimrc

/etc/vimrc:
  file.managed:
    - source: salt://edit/vimrc

So here, with a requisite_in, the same thing is accomplished, but just from the other way around. The vim package is saying "/etc/vimrc depends on me".

In the end, a single dependency map is created and everything is executed in a finite and predictable order.

Note

Requisite matching

Requisites match on both the ID Declaration and the name parameter. This means that, in the example above, the require_in requisite would also have been matched if the /etc/vimrc state was written as follows:

vimrc:
  file.managed:
    - name: /etc/vimrc
    - source: salt://edit/vimrc

Requisite and Requisite in types¶

There are three requisite statements that can be used in Salt. the require, watch and use requisites. Each requisite also has a corresponding requisite_in: require_in, watch_in and use_in. All of the requisites define specific relationships and always work with the dependency logic defined above.

Require¶

The most basic requisite statement is require. The behavior of require is simple. Make sure that the dependent state is executed before the depending state, and if the dependent state fails, don't run the depending state. So in the above examples the file /etc/vimrc will only be applied after the vim package is installed and only if the vim package is installed successfully.

Require an entire sls file¶

As of Salt 0.16.0, it is possible to require an entire sls file. Do this by first including the sls file and then setting a state to require the included sls file.

include:
  - foo

bar:
  pkg.installed:
    - require:
      - sls: foo

Watch¶

The watch statement does everything the require statement does, but with a little more. The watch statement looks into the state modules for a function called mod_watch. If this function is not available in the corresponding state module, then watch does the same thing as require. If the mod_watch function is in the state module, then the watched state is checked to see if it made any changes to the system, if it has, then mod_watch is called.

Perhaps the best example of using watch is with a service.running state. When a service watches a state, then the service is reloaded/restarted when the watched state changes:

ntpd:
  service.running:
    - watch:
      - file: /etc/ntp.conf
  file.managed:
    - name: /etc/ntp.conf
    - source: salt://ntp/files/ntp.conf

Prereq¶

The prereq requisite is a powerful requisite added in 0.16.0. This requisite allows for actions to be taken based on the expected results of a state that has not yet been executed. In more practical terms, a service can be shut down because the prereq knows that underlying code is going to be updated and the service should be off-line while the update occurs.

The motivation to add this requisite was to allow for routines to remove a system from a load balancer while code is being updated.

The prereq checks if the required state expects to have any changes by running the single state with test=True. If the pre-required state returns changes, then the state requiring it will execute.

graceful-down:
  cmd.run:
    - name: service apache graceful
    - prereq:
      - file: site-code

site-code:
  file.recurse:
    - name: /opt/site_code
    - source: salt://site/code

In this case the apache server will only be shutdown if the site-code state expects to deploy fresh code via the file.recurse call, and the site-code deployment will only be executed if the graceful-down run completes successfully.

Use¶

The use requisite is used to inherit the arguments passed in another id declaration. This is useful when many files need to have the same defaults.

The use statement was developed primarily for the networking states but can be used on any states in Salt. This made sense for the networking state because it can define a long list of options that need to be applied to multiple network interfaces.

Require In¶

The require_in requisite is the literal reverse of require. If a state declaration needs to be required by another state declaration then require_in can accommodate it, so these two sls files would be the same in the end:

Using require

httpd:
  pkg:
    - installed
  service:
    - running
    - require:
      - pkg: httpd

Using require_in

httpd:
  pkg:
    - installed
    - require_in:
      - service: httpd
  service:
    - running

The require_in statement is particularly useful when assigning a require in a separate sls file. For instance it may be common for httpd to require components used to set up PHP or mod_python, but the HTTP state does not need to be aware of the additional components that require it when it is set up:

http.sls

httpd:
  pkg:
    - installed
  service:
    - running
    - require:
      - pkg: httpd

php.sls

include:
  - http

php:
  pkg:
    - installed
    - require_in:
      - service: httpd

mod_python.sls

include:
  - http

mod_python:
  pkg:
    - installed
    - require_in:
      - service: httpd

Now the httpd server will only start if php or mod_python are first verified to be installed. Thus allowing for a requisite to be defined "after the fact".

Watch In¶

Watch in functions the same was as require in, but applies a watch statement rather than a require statement to the external state declaration.

Prereq In¶

The prereq_in requisite in follows the same assignment logic as the require_in requisite in. The prereq_in call simply assigns prereq to the state referenced. The above example for prereq can be modified to function in the same way using prereq_in:

graceful-down:
  cmd.run:
    - name: service apache graceful

site-code:
  file.recurse:
    - name: /opt/site_code
    - source: salt://site/code
    - prereq_in:
      - cmd: graceful-down

Startup States¶

Sometimes it may be desired that the salt minion execute a state run when it is started. This alleviates the need for the master to initiate a state run on a new minion and can make provisioning much easier.

As of Salt 0.10.3 the minion config reads options that allow for states to be executed at startup. The options are startup_states, sls_list and top_file.

The startup_states option can be passed one of a number of arguments to define how to execute states. The available options are:

highstate: Execute state.highstate
sls: Read in the sls_list option and execute the named sls files
top: Read in the top_file option and execute states based on that top file on the Salt Master

Examples:¶

Execute state.highstate when starting the minion:

startup_states: highstate

Execute the sls files edit.vim and hyper:

startup_states: sls

sls_list:
  - edit.vim
  - hyper

State Testing¶

Executing a Salt state run can potentially change many aspects of a system and it may be desirable to first see what a state run is going to change before applying the run.

Salt has a test interface to report on exactly what will be changed, this interface can be invoked on any of the major state run functions:

salt '*' state.highstate test=True
salt '*' state.sls test=True
salt '*' state.single test=True

The test run is mandated by adding the test=True option to the states. The return information will show states that will be applied in yellow and the result is reported as None.

Default Test¶

If the value test is set to True in the minion configuration file then states will default to being executed in test mode. If this value is set then states can still be run by calling test=False:

salt '*' state.highstate test=False
salt '*' state.sls test=False
salt '*' state.single test=False

The Top File¶

The top file is used to map what SLS modules get loaded onto what minions via the state system. The top file creates a few general abstractions. First it maps what nodes should pull from which environments, next it defines which matches systems should draw from.

Environments¶

Environment: A configuration that allows conceptually organizing state tree directories. Environments can be made to be self-contained or state trees can be made to bleed through environments.

The environments in the top file corresponds with the environments defined in the file_roots variable. In a simple, single environment setup you only have the base environment, and therefore only one state tree. Here is a simple example of file_roots in the master configuration:

file_roots:
  base:
    - /srv/salt

This means that the top file will only have one environment to pull from, here is a simple, single environment top file:

base:
  '*':
    - core
    - edit

This also means that /srv/salt has a state tree. But if you want to use multiple environments, or partition the file server to serve more than just the state tree, then the file_roots option can be expanded:

file_roots:
  base:
    - /srv/salt/base
  dev:
    - /srv/salt/dev
  qa:
    - /srv/salt/qa
  prod:
    - /srv/salt/prod

Then our top file could reference the environments:

dev:
  'webserver*dev*':
    - webserver
  'db*dev*':
    - db
qa:
  'webserver*qa*':
    - webserver
  'db*qa*':
    - db
prod:
  'webserver*prod*':
    - webserver
  'db*prod*':
    - db

In this setup we have state trees in three of the four environments, and no state tree in the base environment. Notice that the targets for the minions specify environment data. In Salt the master determines who is in what environment, and many environments can be crossed together. For instance, a separate global state tree could be added to the base environment if it suits your deployment:

base:
  '*':
    - global
dev:
  'webserver*dev*':
    - webserver
  'db*dev*':
    - db
qa:
  'webserver*qa*':
    - webserver
  'db*qa*':
    - db
prod:
  'webserver*prod*':
    - webserver
  'db*prod*':
    - db

In this setup all systems will pull the global SLS from the base environment, as well as pull from their respective environments. If you assign only one SLS to a system, as in this example, a shorthand is also available:

base:
  '*': global
dev:
  'webserver*dev*': webserver
  'db*dev*':        db
qa:
  'webserver*qa*': webserver
  'db*qa*':        db
prod:
  'webserver*prod*': webserver
  'db*prod*':        db

Note

The top files from all defined environments will be compiled into a single top file for all states. Top files are environment agnostic.

Remember, that since everything is a file in Salt, the environments are primarily file server environments, this means that environments that have nothing to do with states can be defined and used to distribute other files.

A clean and recommended setup for multiple environments would look like this:

# Master file_roots configuration:
file_roots:
  base:
    - /srv/salt/base
  dev:
    - /srv/salt/dev
  qa:
    - /srv/salt/qa
  prod:
    - /srv/salt/prod

Then only place state trees in the dev, qa and prod environments, leaving the base environment open for generic file transfers. Then the top.sls file would look something like this:

dev:
  'webserver*dev*':
    - webserver
  'db*dev*':
    - db
qa:
  'webserver*qa*':
    - webserver
  'db*qa*':
    - db
prod:
  'webserver*prod*':
    - webserver
  'db*prod*':
    - db

Other Ways of Targeting Minions¶

In addition to globs, minions can be specified in top files a few other ways. Some common ones are compound matches and node groups.

Here is a slightly more complex top file example, showing the different types of matches you can perform:

base:
    '*':
        - ldap-client
        - networking
        - salt.minion

    'salt-master*':
        - salt.master

    '^(memcache|web).(qa|prod).loc$':
        - match: pcre
        - nagios.mon.web
        - apache.server

    'os:Ubuntu':
        - match: grain
        - repos.ubuntu

    'os:(RedHat|CentOS)':
        - match: grain_pcre
        - repos.epel

    'foo,bar,baz':
        - match: list
        - database

    'somekey:abc':
        - match: pillar
        - xyz

    'nag1* or G@role:monitoring':
        - match: compound
        - nagios.server

In this example top.sls, all minions get the ldap-client, networking and salt.minion states. Any minion with an id matching the salt-master* glob will get the salt.master state. Any minion with ids matching the regular expression ^(memcache|web).(qa|prod).loc$ will get the nagios.mon.web and apache.server states. All Ubuntu minions will receive the repos.ubuntu state, while all RHEL and CentOS minions will receive the repos.epel state. The minions foo, bar, and baz will receive the database state. Any minion with a pillar named somekey, having a value of abc will receive the xyz state. Finally, minions with ids matching the nag1* glob or with a grain named role equal to monitoring will receive the nagios.server state.

How Top Files Are Compiled¶

As mentioned earlier, the top files in the different environments are compiled into a single set of data. The way in which this is done follows a few rules, which are important to understand when arranging top files in different environments. The examples below all assume that the file_roots are set as in the above multi-environment example.

The base environment's top file is processed first. Any environment which is defined in the base top.sls as well as another environment's top file, will use the instance of the environment configured in base and ignore all other instances. In other words, the base top file is authoritative when defining environments. Therefore, in the example below, the dev section in /srv/salt/dev/top.sls would be completely ignored.

/srv/salt/base/top.sls:

base:
  '*':
    - common
dev:
  'webserver*dev*':
    - webserver
  'db*dev*':
    - db

/srv/salt/dev/top.sls:

dev:
  '10.10.100.0/24':
    - match: ipcidr
    - deployments.dev.site1
  '10.10.101.0/24':
    - match: ipcidr
    - deployments.dev.site2

Note

The rules below assume that the environments being discussed were not defined in the base top file.

If, for some reason, the base environment is not configured in the base environment's top file, then the other environments will be checked in alphabetical order. The first top file found to contain a section for the base environment wins, and the other top files' base sections are ignored. So, provided there is no base section in the base top file, with the below two top files the dev environment would win out, and the common.centos SLS would not be applied to CentOS hosts.

/srv/salt/dev/top.sls:

base:
  'os:Ubuntu':
    - common.ubuntu
dev:
  'webserver*dev*':
    - webserver
  'db*dev*':
    - db

/srv/salt/qa/topsls:

base:
  'os:Ubuntu':
    - common.ubuntu
  'os:CentOS':
    - common.centos
qa:
  'webserver*qa*':
    - webserver
  'db*qa*':
    - db

For environments other than base, the top file in a given environment will be checked for a section matching the environment's name. If one is found, then it is used. Otherwise, the remaining (non-base) environments will be checked in alphabetical order. In the below example, the qa section in /srv/salt/dev/top.sls will be ignored, but if /srv/salt/qa/top.sls were cleared or removed, then the states configured for the qa environment in /srv/salt/dev/top.sls will be applied.

/srv/salt/dev/top.sls:

dev:
  'webserver*dev*':
    - webserver
  'db*dev*':
    - db
qa:
  '10.10.200.0/24':
    - match: ipcidr
    - deployments.qa.site1
  '10.10.201.0/24':
    - match: ipcidr
    - deployments.qa.site2

/srv/salt/qa/top.sls:

qa:
  'webserver*qa*':
    - webserver
  'db*qa*':
    - db

Note

When in doubt, the simplest way to configure your states is with a single top.sls in the base environment.

SLS Template Variable Reference¶

The template engines available to sls files and file templates come loaded with a number of context variables. These variables contain information and functions to assist in the generation of templates.

Salt¶

The salt variable is available to abstract the salt library functions. This variable is a python dictionary containing all of the functions available to the running salt minion:

{% for file in salt['cmd.run'](ls /opt/to_remove) %}
{{ file }}:
  file.absent
{% endfor %}

Opts¶

The opts variable abstracts the contents of the minion's configuration file directly to the template. The opts variable is a dictionary.

{{ opts['cachedir'] }}

The config.get function also searches for values in the opts dictionary.

Pillar¶

The pillar dictionary can be referenced directly:

{{ pillar['key'] }}

Using the pillar.get function via the salt variable is generally recommended since a default can be safely set in the event that the value is not available in pillar and dictionaries can be traversed directly:

{{ salt['pillar.get']('key', 'failover_value') }}
{{ salt['pillar.get']('stuff:more:deeper') }}

Grains¶

The grains dictionary makes the minion's grains directly available:

{{ grains['os'] }}

The grains.get function can be used to traverse deeper grains and set defaults:

{{ salt['grains.get']('os') }}

env¶

The env variable is available in sls files when gathering the sls from an environment.

{{ env }}

sls¶

The sls variable contains the sls reference value. The sls reference value is the value used to include the sls in top files or via the include option.

{{ sls }}

State Modules¶

State Modules are the components that map to actual enforcement and management of Salt states.

States are Easy to Write!¶

State Modules should be easy to write and straightforward. The information passed to the SLS data structures will map directly to the states modules.

Mapping the information from the SLS data is simple, this example should illustrate:

/etc/salt/master: # maps to "name"
  file: # maps to State module filename e.g. https://github.com/saltstack/salt/blob/develop/salt/states/file.py
    - managed # maps to the managed function in the file State module
    - user: root # one of many options passed to the manage function
    - group: root
    - mode: 644
    - source: salt://salt/master

Therefore this SLS data can be directly linked to a module, function and arguments passed to that function.

This does issue the burden, that function names, state names and function arguments should be very human readable inside state modules, since they directly define the user interface.

Keyword Arguments

Salt passes a number of keyword arguments to states when rendering them, including the environment, a unique identifier for the state, and more. Additionally, keep in mind that the requisites for a state are part of the keyword arguments. Therefore, if you need to iterate through the keyword arguments in a state, these must be considered and handled appropriately. One such example is in the pkgrepo.managed state, which needs to be able to handle arbitrary keyword arguments and pass them to module execution functions. An example of how these keyword arguments can be handled can be found here.

Using Custom State Modules¶

Place your custom state modules inside a _states directory within the file_roots specified by the master config file. These custom state modules can then be distributed in a number of ways. Custom state modules are distributed when state.highstate is run, or by executing the saltutil.sync_states or saltutil.sync_all functions.

Any custom states which have been synced to a minion, that are named the same as one of Salt's default set of states, will take the place of the default state with the same name. Note that a state's default name is its filename (i.e. foo.py becomes state foo), but that its name can be overridden by using a __virtual__ function.

Cross Calling Modules¶

As with Execution Modules, State Modules can also make use of the __salt__ and __grains__ data.

It is important to note that the real work of state management should not be done in the state module unless it is needed. A good example is the pkg state module. This module does not do any package management work, it just calls the pkg execution module. This makes the pkg state module completely generic, which is why there is only one pkg state module and many backend pkg execution modules.

On the other hand some modules will require that the logic be placed in the state module, a good example of this is the file module. But in the vast majority of cases this is not the best approach, and writing specific execution modules to do the backend work will be the optimal solution.

Return Data¶

A State Module must return a dict containing the following keys/values:

name: The same value passed to the state as "name".
changes: A dict describing the changes made. Each thing changed should be a key, with its value being another dict with keys called "old" and "new" containing the old/new values. For example, the pkg state's changes dict has one key for each package changed, with the "old" and "new" keys in its sub-dict containing the old and new versions of the package.
result: A boolean value. True if the action was successful, otherwise False.
comment: A string containing a summary of the result.

Test State¶

All states should check for and support test being passed in the options. This will return data about what changes would occur if the state were actually run. An example of such a check could look like this:

# Return comment of changes if test.
if __opts__['test']:
    ret['result'] = None
    ret['comment'] = 'State Foo will execute with param {0}'.format(bar)
    return ret

Make sure to test and return before performing any real actions on the minion.

Watcher Function¶

If the state being written should support the watch requisite then a watcher function needs to be declared. The watcher function is called whenever the watch requisite is invoked and should be generic to the behavior of the state itself.

The watcher function should accept all of the options that the normal state functions accept (as they will be passed into the watcher function).

A watcher function typically is used to execute state specific reactive behavior, for instance, the watcher for the service module restarts the named service and makes it useful for the watcher to make the service react to changes in the environment.

The watcher function also needs to return the same data that a normal state function returns.

Mod_init Interface¶

Some states need to execute something only once to ensure that an environment has been set up, or certain conditions global to the state behavior can be predefined. This is the realm of the mod_init interface.

A state module can have a function called mod_init which executes when the first state of this type is called. This interface was created primarily to improve the pkg state. When packages are installed the package metadata needs to be refreshed, but refreshing the package metadata every time a package is installed is wasteful. The mod_init function for the pkg state sets a flag down so that the first, and only the first, package installation attempt will refresh the package database (the package database can of course be manually called to refresh via the refresh option in the pkg state).

The mod_init function must accept the Low State Data for the given executing state as an argument. The low state data is a dict and can be seen by executing the state.show_lowstate function. Then the mod_init function must return a bool. If the return value is True, then the mod_init function will not be executed again, meaning that the needed behavior has been set up. Otherwise, if the mod_init function returns False, then the function will be called the next time.

A good example of the mod_init function is found in the pkg state module:

def mod_init(low):
    '''
    Refresh the package database here so that it only needs to happen once
    '''
    if low['fun'] == 'installed' or low['fun'] == 'latest':
        rtag = __gen_rtag()
        if not os.path.exists(rtag):
            open(rtag, 'w+').write('')
        return True
    else:
        return False

The mod_init function in the pkg state accepts the low state data as low and then checks to see if the function being called is going to install packages, if the function is not going to install packages then there is no need to refresh the package database. Therefore if the package database is prepared to refresh, then return True and the mod_init will not be called the next time a pkg state is evaluated, otherwise return False and the mod_init will be called next time a pkg state is evaluated.

Full list of builtin state modules¶

`alias`	Configuration of email aliases.
`alternatives`	Configuration of the alternatives system
`apt`	Package management operations specific to APT- and DEB-based systems
`augeas`	Configuration management using Augeas
`cmd`	Execution of arbitrary commands
`cron`	Management of cron, the Unix command scheduler.
`debconfmod`	Management of debconf selections.
`disk`	Disk monitoring state
`eselect`	Management of Gentoo configuration using eselect
`file`	Operations on regular files, special files, directories, and symlinks.
`gem`	Installation of Ruby modules packaged as gems.
`git`	Interaction with Git repositories.
`grains`	Manage grains on the minion.
`group`	Management of user groups.
`hg`	Interaction with Mercurial repositories.
`host`	Management of addresses and names in hosts file.
`iptables`	Management of iptables
`keyboard`	Management of keyboard layouts
`kmod`	Loading and unloading of kernel modules.
`layman`	Management of Gentoo Overlays using layman
`libvirt`	Manage libvirt certs.
`locale`	Management of languages/locales
`lvm`	Management of Linux logical volumes
`makeconf`	Management of Gentoo make.conf
`mdadm`	Managing software RAID with mdadm
`modjk_worker`	Send commands to a modjk load balancer via the peer system
`module`	Execution of Salt modules from within states.
`mongodb_database`	Management of Mongodb databases
`mongodb_user`	Management of Mongodb users
`mount`	Mounting of filesystems.
`mysql_database`	Management of MySQL databases (schemas).
`mysql_grants`	Management of MySQL grants (user permissions).
`mysql_user`	Management of MySQL users.
`network`	Configuration of network interfaces.
`npm`	Installation of NPM Packages
`pecl`	Installation of PHP Extensions Using pecl
`pip_state`	Installation of Python Packages Using pip
`pkg`	Installation of packages using OS package managers such as yum or apt-get
`pkgng`	Manage package remote repo using FreeBSD pkgng
`pkgrepo`	Management of package repos
`portage_config`	Management of Portage package configuration on Gentoo
`postgres_database`	Management of PostgreSQL databases.
`postgres_group`	Management of PostgreSQL groups (roles).
`postgres_user`	Management of PostgreSQL users (roles).
`quota`	Management of POSIX Quotas
`rabbitmq_user`	Manage RabbitMQ Users.
`rabbitmq_vhost`	Manage RabbitMQ Virtual Hosts.
`rbenv`	Managing Ruby installations with rbenv.
`rvm`	Managing Ruby installations and gemsets with Ruby Version Manager (RVM).
`selinux`	Management of SELinux rules.
`service`	Starting or restarting of services and daemons.
`ssh_auth`	Control of entries in SSH authorized_key files.
`ssh_known_hosts`	Control of SSH known_hosts entries.
`stateconf`	Stateconf System
`supervisord`	Interaction with the Supervisor daemon.
`svn`	Manage SVN repositories
`sysctl`	Configuration of the Linux kernel using sysctrl.
`timezone`	Management of timezones
`tomcat`	This state uses the manager webapp to manage Apache tomcat webapps
`user`	Management of user accounts.
`virtualenv_mod`	Setup of Python virtualenv sandboxes.

Renderers¶

The Salt state system operates by gathering information from simple data structures. The state system was designed in this way to make interacting with it generic and simple. This also means that state files (SLS files) can be one of many formats.

By default SLS files are rendered as Jinja templates and then parsed as YAML documents. But since the only thing the state system cares about is raw data, the SLS files can be any structured format that can be dreamed up.

Currently there is support for Jinja + YAML, Mako + YAML, Wempy + YAML, Jinja + json Mako + json and Wempy + json. But renderers can be written to support anything. This means that the Salt states could be managed by XML files, HTML files, puppet files, or any format that can be translated into the data structure used by the state system.

Multiple Renderers¶

When deploying a state tree a default renderer is selected in the master configuration file with the renderer option. But multiple renderers can be used inside the same state tree.

When rendering SLS files Salt checks for the presence of a Salt specific shebang line. The shebang line syntax was chosen because it is familiar to the target audience, the systems admin and systems engineer.

The shebang line directly calls the name of the renderer as it is specified within Salt. One of the most common reasons to use multiple renderers in to use the Python or py renderer:

#!py

def run():
    '''
    Install the python-mako package
    '''
    return {'include': ['python'],
            'python-mako': {'pkg': ['installed']}}

The first line is a shebang that references the py renderer.

Composing Renderers¶

A renderer can be composed from other renderers by connecting them in a series of pipes(|). In fact, the default Jinja + YAML renderer is implemented by combining a YAML renderer and a Jinja renderer. Such renderer configuration is specified as: jinja | yaml.

Other renderer combinations are possible, here's a few examples:

yaml

i.e, just YAML, no templating.

mako | yaml

pass the input to the mako renderer, whose output is then fed into the yaml renderer.

jinja | mako | yaml

This one allows you to use both jinja and mako templating syntax in the input and then parse the final rendered output as YAML.

And here's a contrived example sls file using the jinja | mako | yaml renderer:

#!jinja|mako|yaml

An_Example:
  cmd.run:
    - name: |
        echo "Using Salt ${grains['saltversion']}" \
             "from path {{grains['saltpath']}}."
    - cwd: /

<%doc> ${...} is Mako's notation, and so is this comment. </%doc>
{#     Similarly, {{...}} is Jinja's notation, and so is this comment. #}

For backward compatibility, jinja | yaml can also be written as yaml_jinja, and similarly, the yaml_mako, yaml_wempy, json_jinja, json_mako, and json_wempy renderers are all supported as well.

Keep in mind that not all renderers can be used alone or with any other renderers. For example, the template renderers shouldn't be used alone as their outputs are just strings, which still need to be parsed by another renderer to turn them into highstate data structures. Also, for example, it doesn't make sense to specify yaml | jinja either, because the output of the yaml renderer is a highstate data structure(a dict in Python), which cannot be used as the input to a template renderer. Therefore, when combining renderers, you should know what each renderer accepts as input and what it returns as output.

Writing Renderers¶

Writing a renderer is easy, all that is required is that a Python module is placed in the rendered directory and that the module implements the render function. The render function will be passed the path of the SLS file. In the render function, parse the passed file and return the data structure derived from the file. You can place your custom renderers in a _renderers directory within the file_roots specified by the master config file. These custom renderers are distributed when state.highstate is run, or by executing the saltutil.sync_renderers or saltutil.sync_all functions.

Any custom renderers which have been synced to a minion, that are named the same as one of Salt's default set of renderers, will take the place of the default renderer with the same name.

Examples¶

The best place to find examples of renderers is in the Salt source code. The renderers included with Salt can be found here:

https://github.com/saltstack/salt/blob/develop/salt/renderers

Here is a simple YAML renderer example:

import yaml
def render(yaml_data, env='', sls='', **kws):
    if not isinstance(yaml_data, basestring):
        yaml_data = yaml_data.read()
    data = yaml.load(yaml_data)
    return data if data else {}

Full list of builtin renderer modules¶

`jinja`
`json`
`mako`
`py`	Pure python state renderer
`pydsl`	A Python-based DSL
`stateconf`	A flexible renderer that takes a templating engine and a data format
`wempy`
`yaml`

Pillars¶

Salt includes a number of built-in external pillars, listed at Full list of builtin pillar modules.

You may also wish to look at the standard pillar documentation, at Pillar Configuration

The source for the built-in Salt pillars can be found here: https://github.com/saltstack/salt/blob/develop/salt/pillar

Full list of builtin pillar modules¶

`cmd_json`	Execute a command and read the output as JSON.
`cmd_yaml`	Execute a command and read the output as YAML.
`cobbler`	Cobbler Pillar ============== A pillar module to pull data from Cobbler via its API into the pillar dictionary.
`django_orm`	Generate pillar data from Django models through the Django ORM
`git_pillar`	Clone a remote git repository and use the filesystem as a pillar directory.
`hiera`	Take in a hiera configuration file location and execute it.
`libvirt`	Load up the libvirt keys into pillar for a given minion if said keys have been generated using the libvirt key runner.
`mongo`	Read pillar data from a mongodb collection.
`pillar_ldap`	This pillar module parses a config file (specified in the salt master config), and executes a series of LDAP searches based on that config.
`puppet`	Execute an unmodified puppet_node_classifier and read the output as YAML.
`reclass_adapter`

Master Tops¶

Salt includes a number of built-in subsystems to generate top file data, they are listed listed at Full list of builtin master tops modules.

The source for the built-in Salt master tops can be found here: https://github.com/saltstack/salt/blob/develop/salt/tops

Full list of builtin master tops modules¶

`cobbler`	Cobbler Tops
`ext_nodes`	External Nodes Classifier
`mongo`	Read tops data from a mongodb collection.
`reclass_adapter`

Salt Runners¶

See also

The full list of runners

Salt runners are convenience applications executed with the salt-run command. Where as salt modules are sent out to minions for execution, salt runners are executed on the salt master.

A Salt runner can be a simple client call, or a complex application.

The use for a Salt runner is to build a frontend hook for running sets of commands via Salt or creating special formatted output.

Writing Salt Runners¶

Salt runners can be easily written, the work in a similar way to Salt modules except they run on the server side.

A runner is a Python module that contains functions, each public function is a runner that can be executed via the salt-run command.

If a Python module named test.py is created in the runners directory and contains a function called foo then the function could be called with:

# salt-run test.foo

Examples¶

The best examples of runners can be found in the Salt source:

https://github.com/saltstack/salt/blob/develop/salt/runners

A simple runner that returns a well-formatted list of the minions that are responding to Salt calls would look like this:

# Import salt modules
import salt.client

def up():
    '''
    Print a list of all of the minions that are up
    '''
    client = salt.client.LocalClient(__opts__['conf_file'])
    minions = client.cmd('*', 'test.ping', timeout=1)
    for minion in sorted(minions):
        print minion

Full list of runner modules¶

`cache`	Return cached data from minions
`doc`	A runner module to collect and display the inline documentation from the
`fileserver`	Directly manage the salt fileserver plugins
`jobs`	A convenience system to manage jobs, both active and already run
`launchd`
`manage`	General management functions for salt, tools like seeing what hosts are up
`network`	Network tools to run from the Master
`search`	Runner frontend to search system
`state`	Execute overstate functions
`virt`	Control virtual machines via Salt
`winrepo`	Runner to manage Windows software repo

Full list of builtin wheel modules¶

`config`	Manage the master configuration file
`file_roots`	Read in files from the file_root and save files to the file root
`key`	Wheel system wrapper for key system
`pillar_roots`	The pillar_roots wheel module is used to manage files under the pillar roots directories on the master server.

Full list of builtin auth modules¶

`keystone`	Provide authentication using OpenStack Keystone
`ldap`	Provide authentication using simple LDAP binds
`pam`	Authenticate against PAM
`stormpath_mod`	Salt Stormpath Authentication

Full list of builtin output modules¶

`grains`	Special outputter for grains
`highstate`	The return data from the Highstate command is a standard data structure which is parsed by the highstate outputter to deliver a clean and readable set of information about the HighState run on minions.
`json_out`	The JSON output module converts the return data into JSON.
`key`	Salt Key makes use of the outputter system to format information sent to the `salt-key` command.
`nested`	Recursively display nested data, this is the default outputter.
`no_out`	Display no output.
`no_return`	Display output for minions that did not return
`overstatestage`	Display clean output of an overstate stage
`pprint_out`	The python pretty print system was the default outputter.
`raw`	The raw outputter outputs the data via the python print function and is shown in a raw state.
`txt`	The txt outputter has been developed to make the output from shell commands on minions appear as they do when the command is executed on the minion.
`virt_query`	virt.query outputter
`yaml_out`	Output data in YAML, this outputter defaults to printing in YAML block mode for better readability.

Python client API¶

Salt is written to be completely API centric, Salt minions and master can be built directly into third party applications as a communication layer. The Salt client API is very straightforward.

A number of client command methods are available depending on the exact behavior desired.

LocalClient¶

class salt.client.LocalClient(c_path='/etc/salt/master', mopts=None)¶

LocalClient is the same interface used by the salt command-line tool on the Salt Master. LocalClient is used to send a command to Salt minions to execute execution modules and return the results to the Salt Master.

Importing and using LocalClient must be done on the same machine as the Salt Master and it must be done using the same user that the Salt Master is running as (unless external_auth is configured and authentication credentials are included in the execution.

cmd(tgt, fun, arg=(), timeout=None, expr_form='glob', ret='', kwarg=None, **kwargs)¶

The cmd method will execute and wait for the timeout period for all minions to reply, then it will return all minion data at once.

Usage:

import salt.client
client = salt.client.LocalClient()
ret = client.cmd('*', 'cmd.run', ['whoami'])

With authentication:

# Master config
...
external_auth:
  pam:
    fred:
      - test.*
...

ret = client.cmd('*', 'test.ping', [], username='fred', password='pw', eauth='pam')

Compound command usage:

ret = client.cmd('*', ['grains.items', 'cmd.run'], [[], ['whoami']])

Parameters:

tgt (string or list) -- Which minions to target for the execution. Default is shell glob. Modified by the expr_form option.
fun (string or list of strings) --
The module and function to call on the specified minions of the form module.function. For example test.ping or grains.items.

Compound commands

Multiple functions may be called in a single publish by passing a list of commands. This can dramatically lower overhead and speed up the application communicating with Salt.
This requires that the arg param is a list of lists. The fun list and the arg list must correlate by index meaning a function that does not take arguments must still have a corresponding empty list at the expected index.
arg (list or list-of-lists) -- A list of arguments to pass to the remote function. If the function takes no arguments arg may be omitted except when executing a compound command.
timeout -- Seconds to wait after the last minion returns but before all minions return.
expr_form --
The type of tgt. Allowed values:
- glob - Bash glob completion - Default
- pcre - Perl style regular expression
- list - Python list of hosts
- grain - Match based on a grain comparison
- grain_pcre - Grain comparison with a regex
- pillar - Pillar data comparison
- nodegroup - Match on nodegroup
- range - Use a Range server for matching
- compound - Pass a compound match string
ret -- The returner to use. The value passed can be single returner, or a comma delimited list of returners to call in order on the minions
kwargs --
Optional keyword arguments.

Authentication credentials may be passed when using external_auth.
- eauth - the external_auth backend
- username and password
- token

Returns:

A dictionary with the result of the execution, keyed by minion ID. A compound command will return a sub-dictionary keyed by function name.

cmd_async(tgt, fun, arg=(), expr_form='glob', ret='', kwarg=None, **kwargs)¶

Execute a command and get back the jid, don't wait for anything

The function signature is the same as cmd() with the following exceptions.

Returns:	A job ID

cmd_cli(tgt, fun, arg=(), timeout=None, expr_form='glob', ret='', verbose=False, kwarg=None, **kwargs)¶

Used by the salt CLI. This method returns minion returns as the come back and attempts to block until all minions return.

The function signature is the same as cmd() with the following exceptions.

Parameters:	verbose -- Print extra information about the running command
Returns:	A generator

cmd_iter(tgt, fun, arg=(), timeout=None, expr_form='glob', ret='', kwarg=None, **kwargs)¶

Yields the individual minion returns as they come in

The function signature is the same as cmd() with the following exceptions.

cmd_iter_no_block(tgt, fun, arg=(), timeout=None, expr_form='glob', ret='', kwarg=None, **kwargs)¶

Blocks while waiting for individual minions to return.

The function signature is the same as cmd() with the following exceptions.

Returns:	None until the next minion returns. This allows for actions to be injected in between minion returns.

Salt Caller¶

class salt.client.Caller(c_path='/etc/salt/minion')¶

Caller is the same interface used by the salt-call command-line tool on the Salt Minion.

Importing and using LocalClient must be done on the same machine as a Salt Minion and it must be done using the same user that the Salt Minion is running as.

Usage:

import salt.client
caller = salt.client.Caller()
caller.function('test.ping')

# Or call objects directly
caller.sminion.functions['cmd.run']('ls -l')

function(fun, *args, **kwargs)¶: Call a single salt function

RunnerClient¶

class salt.runner.RunnerClient(opts)¶

RunnerClient is the same interface used by the salt-run command-line tool on the Salt Master. It executes runner modules which run on the Salt Master.

Importing and using RunnerClient must be done on the same machine as the Salt Master and it must be done using the same user that the Salt Master is running as.

cmd(fun, arg, kwarg=None)¶: Execute a runner with the given arguments

low(fun, low)¶: Pass in the runner function name and the low data structure

WheelClient¶

class salt.wheel.Wheel(opts)¶

WheelClient is an interface to Salt's wheel modules. Wheel modules interact with various parts of the Salt Master.

Importing and using WheelClient must be done on the same machine as the Salt Master and it must be done using the same user that the Salt Master is running as.

call_func(fun, **kwargs)¶: Execute a master control function

master_call(**kwargs)¶: Send a function call to a wheel module through the master network interface Expects that one of the kwargs is key 'fun' whose value is the namestring of the function to call

Peer Communication¶

Salt 0.9.0 introduced the capability for Salt minions to publish commands. The intent of this feature is not for Salt minions to act as independent brokers one with another, but to allow Salt minions to pass commands to each other.

In Salt 0.10.0 the ability to execute runners from the master was added. This allows for the master to return collective data from runners back to the minions via the peer interface.

The peer interface is configured through two options in the master configuration file. For minions to send commands from the master the peer configuration is used. To allow for minions to execute runners from the master the peer_run configuration is used.

Since this presents a viable security risk by allowing minions access to the master publisher the capability is turned off by default. The minions can be allowed access to the master publisher on a per minion basis based on regular expressions. Minions with specific ids can be allowed access to certain Salt modules and functions.

Peer Configuration¶

The configuration is done under the peer setting in the Salt master configuration file, here are a number of configuration possibilities.

The simplest approach is to enable all communication for all minions, this is only recommended for very secure environments.

peer:
  .*:
    - .*

This configuration will allow minions with IDs ending in example.com access to the test, ps, and pkg module functions.

peer:
  .*example.com:
    - test.*
    - ps.*
    - pkg.*

The configuration logic is simple, a regular expression is passed for matching minion ids, and then a list of expressions matching minion functions is associated with the named minion. For instance, this configuration will also allow minions ending with foo.org access to the publisher.

peer:
  .*example.com:
    - test.*
    - ps.*
    - pkg.*
  .*foo.org:
    - test.*
    - ps.*
    - pkg.*

Peer Runner Communication¶

Configuration to allow minions to execute runners from the master is done via the peer_run option on the master. The peer_run configuration follows the same logic as the peer option. The only difference is that access is granted to runner modules.

To open up access to all minions to all runners:

peer_run:
  .*:
    - .*

This configuration will allow minions with IDs ending in example.com access to the manage and jobs runner functions.

peer_run:
  .*example.com:
    - manage.*
    - jobs.*

Using Peer Communication¶

The publish module was created to manage peer communication. The publish module comes with a number of functions to execute peer communication in different ways. Currently there are three functions in the publish module. These examples will show how to test the peer system via the salt-call command.

To execute test.ping on all minions:

# salt-call publish.publish \* test.ping

To execute the manage.up runner:

# salt-call publish.runner manage.up

To match minions using other matchers, use expr_form:

# salt-call publish.publish 'webserv* and not G@os:Ubuntu' test.ping expr_form='compound'

Client ACL system¶

The salt client ACL system is a means to allow system users other than root to have access to execute select salt commands on minions from the master.

The client ACL system is configured in the master configuration file via the client_acl configuration option. Under the client_acl configuration option the users open to send commands are specified and then a list of regular expressions which specify the minion functions which will be made available to specified user. This configuration is much like the peer configuration:

# Allow thatch to execute anything and allow fred to use ping and pkg
client_acl:
  thatch:
    - .*
  fred:
    - ping.*
    - pkg.*

Permission Issues¶

Directories required for client_acl must be modified to be readable by the users specified:

chmod 755 /var/cache/salt /var/cache/salt/jobs /var/run/salt

If you are upgrading from earlier versions of salt you must also remove any existing user keys and re-start the Salt master:

rm /var/cache/salt/.*key
service salt-master restart

Salt Syndic¶

The Salt Syndic interface is a powerful tool which allows for the construction of Salt command topologies. A basic Salt setup has a Salt Master commanding a group of Salt Minions. The Syndic interface is a special passthrough minion, it is run on a master and connects to another master, then the master that the Syndic minion is listening to can control the minions attached to the master running the syndic.

The intent for supporting many layouts is not presented with the intent of supposing the use of any single topology, but to allow a more flexible method of controlling many systems.

Configuring the Syndic¶

Since the Syndic only needs to be attached to a higher level master the configuration is very simple. On a master that is running a syndic to connect to a higher level master the syndic_master option needs to be set in the master config file. The syndic_master option contains the hostname or IP address of the master server that can control the master that the syndic is running on.

The master that the syndic connects to sees the syndic as an ordinary minion, and treats it as such. the higher level master will need to accept the syndic's minion key like any other minion. This master will also need to set the order_masters value in the configuration to True. The order_masters option in the config on the higher level master is very important, to control a syndic extra information needs to be sent with the publications, the order_masters option makes sure that the extra data is sent out.

To sum up, you have those configuration options available on the master side:

syndic_master: MasterOfMaster ip/address

syndic_master_port: MasterOfMaster ret_port

syndic_log_file: path to the logfile (absolute or not)

syndic_pidfile: path to the pidfile (absolute or not)

Running the Syndic¶

The Syndic is a separate daemon that needs to be started on the master that is controlled by a higher master. Starting the Syndic daemon is the same as starting the other Salt daemons.

# salt-syndic

File Server Backends¶

Salt version 0.12.0 introduced the ability for the Salt Master to integrate different file server backends. File server backends allows the Salt file server to act as a transparent bridge to external resources. The primary example of this is the git backend which allows for all of the Salt formulas and files to be maintained in a remote git repository.

The fileserver backend system can accept multiple backends as well. This makes it possible to have the environments listed in the file_roots configuration available in addition to other backends, or the ability to mix multiple backends.

This feature is managed by the fileserver_backend option in the master config. The desired backend systems are listed in order of search priority:

fileserver_backend:
  - roots
  - git

If this configuration the environments and files defined in the file_roots configuration will be searched first, if the referenced environment and file is not found then the git backend will be searched.

Environments¶

The concept of environments is followed in all backend systems. The environments in the classic roots backend are defined in the file_roots option. Environments map differently based on the backend, for instance the git backend translated branches and tags in git to environments. This makes it easy to define environments in git by just setting a tag or forking a branch.

Dynamic Module Distribution¶

New in version 0.9.5.

Salt Python modules can be distributed automatically via the Salt file server. Under the root of any environment defined via the file_roots option on the master server directories corresponding to the type of module can be used.

Module sync: Automatically transfer and load modules, grains, renderers, returners, states, etc from the master to the minions.

The directories are prepended with an underscore:

_modules

_grains

_renderers

_returners

_states

The contents of these directories need to be synced over to the minions after Python modules have been created in them. There are a number of ways to sync the modules.

Sync Via States¶

The minion configuration contains an option autoload_dynamic_modules which defaults to True. This option makes the state system refresh all dynamic modules when states are run. To disable this behavior set autoload_dynamic_modules to False in the minion config.

When dynamic modules are autoloaded via states, modules only pertinent to the environments matched in the master's top file are downloaded.

This is important to remember, because modules can be manually loaded from any specific environment that environment specific modules will be loaded when a state run is executed.

Sync Via the saltutil Module¶

The saltutil module has a number of functions that can be used to sync all or specific dynamic modules. The saltutil module function saltutil.sync_all will sync all module types over to a minion. For more information see: salt.modules.saltutil

File Server Configuration¶

The Salt file server is a high performance file server written in ZeroMQ. It manages large files quickly and with little overhead, and has been optimized to handle small files in an extremely efficient manner.

The Salt file server is an environment aware file server. This means that files can be allocated within many root directories and accessed by specifying both the file path and the environment to search. The individual environments can span across multiple directory roots to create overlays and to allow for files to be organized in many flexible ways.

Environments¶

The Salt file server defaults to the mandatory base environment. This environment MUST be defined and is used to download files when no environment is specified.

Environments allow for files and sls data to be logically separated, but environments are not isolated from each other. This allows for logical isolation of environments by the engineer using Salt, but also allows for information to be used in multiple environments.

Directory Overlay¶

The environment setting is a list of directories to publish files from. These directories are searched in order to find the specified file and the first file found is returned.

This means that directory data is prioritized based on the order in which they are listed. In the case of this file_roots configuration:

file_roots:
  base:
    - /srv/salt/base
    - /srv/salt/failover

If a file's URI is salt://httpd/httpd.conf, it will first search for the file at /srv/salt/base/httpd/httpd.conf. If the file is found there it will be returned. If the file is not found there, then /srv/salt/failover/httpd/httpd.conf will be used for the source.

This allows for directories to be overlaid and prioritized based on the order they are defined in the configuration.

Local File Server¶

New in version 0.9.8.

The file server can be rerouted to run from the minion. This is primarily to enable running Salt states without a Salt master. To use the local file server interface, copy the file server data to the minion and set the file_roots option on the minion to point to the directories copied from the master. Once the minion file_roots option has been set, change the file_client option to local to make sure that the local file server interface is used.

Salt File Server¶

Salt comes with a simple file server suitable for distributing files to the Salt minions. The file server is a stateless ZeroMQ server that is built into the Salt master.

The main intent of the Salt file server is to present files for use in the Salt state system. With this said, the Salt file server can be used for any general file transfer from the master to the minions.

The cp Module¶

The cp module is the home of minion side file server operations. The cp module is used by the Salt state system, salt-cp and can be used to distribute files presented by the Salt file server.

Environments¶

Since the file server is made to work with the Salt state system, it supports environments. The environments are defined in the master config file and when referencing an environment the file specified will be based on the root directory of the environment.

get_file¶

The cp.get_file function can be used on the minion to download a file from the master, the syntax looks like this:

# salt '*' cp.get_file salt://vimrc /etc/vimrc

This will instruct all Salt minions to download the vimrc file and copy it to /etc/vimrc

Template rendering can be enabled on both the source and destination file names like so:

# salt '*' cp.get_file "salt://{{grains.os}}/vimrc" /etc/vimrc template=jinja

This example would instruct all Salt minions to download the vimrc from a directory with the same name as their OS grain and copy it to /etc/vimrc

For larger files, the cp.get_file module also supports gzip compression. Because gzip is CPU-intensive, this should only be used in scenarios where the compression ratio is very high (e.g. pretty-printed JSON or YAML files).

Use the gzip named argument to enable it. Valid values are 1..9, where 1 is the lightest compression and 9 the heaviest. 1 uses the least CPU on the master (and minion), 9 uses the most.

# salt '*' cp.get_file salt://vimrc /etc/vimrc gzip=5

Finally, note that by default cp.get_file does not create new destination directories if they do not exist. To change this, use the makedirs argument:

# salt '*' cp.get_file salt://vimrc /etc/vim/vimrc makedirs=True

In this example, /etc/vim/ would be created if it didn't already exist.

get_dir¶

The cp.get_dir function can be used on the minion to download an entire directory from the master. The syntax is very similar to get_file:

# salt '*' cp.get_dir salt://etc/apache2 /etc

cp.get_dir supports template rendering and gzip compression arguments just like get_file:

# salt '*' cp.get_dir salt://etc/{{pillar.webserver}} /etc gzip=5 template=jinja

File Server Client API¶

A client API is available which allows for modules and applications to be written which make use of the Salt file server.

The file server uses the same authentication and encryption used by the rest of the Salt system for network communication.

FileClient Class¶

The FileClient class is used to set up the communication from the minion to the master. When creating a FileClient object the minion configuration needs to be passed in. When using the FileClient from within a minion module the built in __opts__ data can be passed:

import salt.minion

def get_file(path, dest, env='base'):
    '''
    Used to get a single file from the Salt master

    CLI Example:
    salt '*' cp.get_file salt://vimrc /etc/vimrc
    '''
    # Create the FileClient object
    client = salt.minion.FileClient(__opts__)
    # Call get_file
    return client.get_file(path, dest, False, env)

Using the FileClient class outside of a minion module where the __opts__ data is not available, it needs to be generated:

import salt.minion
import salt.config

def get_file(path, dest, env='base'):
    '''
    Used to get a single file from the Salt master
    '''
    # Get the configuration data
    opts = salt.config.minion_config('/etc/salt/minion')
    # Create the FileClient object
    client = salt.minion.FileClient(opts)
    # Call get_file
    return client.get_file(path, dest, False, env)

Full list of builtin fileserver modules¶

`gitfs`	The backend for the git based file server system.
`hgfs`	The backed for the mercurial based file server system.
`roots`	The default file server backend
`s3fs`	The backend for a fileserver based on Amazon S3

Example master configuration file ¶

##### Primary configuration settings #####
##########################################
# This configuration file is used to manage the behavior of the Salt Master
# Values that are commented out but have no space after the comment are
# defaults that need not be set in the config. If there is a space after the
# comment that the value is presented as an example and is not the default.

# Per default, the master will automatically include all config files
# from master.d/*.conf (master.d is a directory in the same directory
# as the main master config file)
#default_include: master.d/*.conf

# The address of the interface to bind to
#interface: 0.0.0.0

# Whether the master should listen for IPv6 connections. If this is set to True,
# the interface option must be adjusted too (for example: "interface: '::'")
#ipv6: False

# The tcp port used by the publisher
#publish_port: 4505

# The user to run the salt-master as. Salt will update all permissions to
# allow the specified user to run the master. If the modified files cause
# conflicts set verify_env to False.
#user: root

# Max open files
# Each minion connecting to the master uses AT LEAST one file descriptor, the
# master subscription connection. If enough minions connect you might start
# seeing on the console(and then salt-master crashes):
#   Too many open files (tcp_listener.cpp:335)
#   Aborted (core dumped)
#
# By default this value will be the one of `ulimit -Hn`, ie, the hard limit for
# max open files.
#
# If you wish to set a different value than the default one, uncomment and
# configure this setting. Remember that this value CANNOT be higher than the
# hard limit. Raising the hard limit depends on your OS and/or distribution,
# a good way to find the limit is to search the internet for(for example):
#   raise max open files hard limit debian
#
#max_open_files: 100000

# The number of worker threads to start, these threads are used to manage
# return calls made from minions to the master, if the master seems to be
# running slowly, increase the number of threads
#worker_threads: 5

# The port used by the communication interface. The ret (return) port is the
# interface used for the file server, authentication, job returnes, etc.
#ret_port: 4506

# Specify the location of the daemon process ID file
#pidfile: /var/run/salt-master.pid

# The root directory prepended to these options: pki_dir, cachedir,
# sock_dir, log_file, autosign_file, extension_modules, key_logfile, pidfile.
#root_dir: /

# Directory used to store public key data
#pki_dir: /etc/salt/pki/master

# Directory to store job and cache data
#cachedir: /var/cache/salt/master

# Verify and set permissions on configuration directories at startup
#verify_env: True

# Set the number of hours to keep old job information in the job cache
#keep_jobs: 24

# Set the default timeout for the salt command and api, the default is 5
# seconds
#timeout: 5

# The loop_interval option controls the seconds for the master's maintinance
# process check cycle. This process updates file server backends, cleans the
# job cache and executes the scheduler.
#loop_interval: 60

# Set the default outputter used by the salt command. The default is "nested"
#output: nested

# By default output is colored, to disable colored output set the color value
# to False
#color: True

# Set the directory used to hold unix sockets
#sock_dir: /var/run/salt/master

# The master maintains a job cache, while this is a great addition it can be
# a burden on the master for larger deployments (over 5000 minions).
# Disabling the job cache will make previously executed jobs unavailable to
# the jobs system and is not generally recommended.
#
#job_cache: True

# Cache minion grains and pillar data in the cachedir.
#minion_data_cache: True

# The master can include configuration from other files. To enable this,
# pass a list of paths to this option. The paths can be either relative or
# absolute; if relative, they are considered to be relative to the directory
# the main master configuration file lives in (this file). Paths can make use
# of shell-style globbing. If no files are matched by a path passed to this
# option then the master will log a warning message.
#
#
# Include a config file from some other path:
#include: /etc/salt/extra_config
#
# Include config from several files and directories:
#include:
#  - /etc/salt/extra_config


#####        Security settings       #####
##########################################
# Enable "open mode", this mode still maintains encryption, but turns off
# authentication, this is only intended for highly secure environments or for
# the situation where your keys end up in a bad state. If you run in open mode
# you do so at your own risk!
#open_mode: False

# Enable auto_accept, this setting will automatically accept all incoming
# public keys from the minions. Note that this is insecure.
#auto_accept: False

# If the autosign_file is specified only incoming keys specified in
# the autosign_file will be automatically accepted. This is insecure.
# Regular expressions as well as globing lines are supported.
#autosign_file: /etc/salt/autosign.conf

# Enable permissive access to the salt keys.  This allows you to run the
# master or minion as root, but have a non-root group be given access to
# your pki_dir.  To make the access explicit, root must belong to the group
# you've given access to.  This is potentially quite insecure.
# If an autosign_file is specified, enabling permissive_pki_access will allow group access
# to that specific file.
#permissive_pki_access: False

# Allow users on the master access to execute specific commands on minions.
# This setting should be treated with care since it opens up execution
# capabilities to non root users. By default this capability is completely
# disabled.
#
#client_acl:
#  larry:
#    - test.ping
#    - network.*
#

# Blacklist any of the following users or modules
#
# This example would blacklist all non sudo users, including root from
# running any commands. It would also blacklist any use of the "cmd"
# module.
# This is completely disabled by default.
#
#client_acl_blacklist:
#  users:
#    - root
#    - '^(?!sudo_).*$'   #  all non sudo users
#  modules:
#    - cmd

# The external auth system uses the Salt auth modules to authenticate and
# validate users to access areas of the Salt system.
#
#external_auth:
#  pam:
#    fred:
#      - test.*
#

# Time (in seconds) for a newly generated token to live. Default: 12 hours
#token_expire: 43200

# Allow minions to push files to the master. This is disabled by default, for
# security purposes.
#file_recv: False 

#####    Master Module Management    #####
##########################################
# Manage how master side modules are loaded

# Add any additional locations to look for master runners
#runner_dirs: []

# Enable Cython for master side modules
#cython_enable: False


#####      State System settings     #####
##########################################
# The state system uses a "top" file to tell the minions what environment to
# use and what modules to use. The state_top file is defined relative to the
# root of the base environment as defined in "File Server settings" below.
#state_top: top.sls

# The master_tops option replaces the external_nodes option by creating
# a plugable system for the generation of external top data. The external_nodes
# option is deprecated by the master_tops option.
# To gain the capabilities of the classic external_nodes system, use the
# following configuration:
# master_tops:
#   ext_nodes: <Shell command which returns yaml>
#
#master_tops: {}

# The external_nodes option allows Salt to gather data that would normally be
# placed in a top file. The external_nodes option is the executable that will
# return the ENC data. Remember that Salt will look for external nodes AND top
# files and combine the results if both are enabled!
#external_nodes: None

# The renderer to use on the minions to render the state data
#renderer: yaml_jinja

# The failhard option tells the minions to stop immediately after the first
# failure detected in the state execution, defaults to False
#failhard: False

# The state_verbose and state_output settings can be used to change the way
# state system data is printed to the display. By default all data is printed.
# The state_verbose setting can be set to True or False, when set to False
# all data that has a result of True and no changes will be suppressed.
#state_verbose: True

# The state_output setting changes if the output is the full multi line
# output for each changed state if set to 'full', but if set to 'terse'
# the output will be shortened to a single line.  If set to 'mixed', the output
# will be terse unless a state failed, in which case that output will be full.
#state_output: full


#####      File Server settings      #####
##########################################
# Salt runs a lightweight file server written in zeromq to deliver files to
# minions. This file server is built into the master daemon and does not
# require a dedicated port.

# The file server works on environments passed to the master, each environment
# can have multiple root directories, the subdirectories in the multiple file
# roots cannot match, otherwise the downloaded files will not be able to be
# reliably ensured. A base environment is required to house the top file.
# Example:
# file_roots:
#   base:
#     - /srv/salt/
#   dev:
#     - /srv/salt/dev/services
#     - /srv/salt/dev/states
#   prod:
#     - /srv/salt/prod/services
#     - /srv/salt/prod/states

#file_roots:
#  base:
#    - /srv/salt

# The hash_type is the hash to use when discovering the hash of a file on
# the master server. The default is md5, but sha1, sha224, sha256, sha384
# and sha512 are also supported.
#hash_type: md5

# The buffer size in the file server can be adjusted here:
#file_buffer_size: 1048576

# A regular expression (or a list of expressions) that will be matched
# against the file path before syncing the modules and states to the minions.
# This includes files affected by the file.recurse state.
# For example, if you manage your custom modules and states in subversion
# and don't want all the '.svn' folders and content synced to your minions,
# you could set this to '/\.svn($|/)'. By default nothing is ignored.
#
#file_ignore_regex:
#  - '/\.svn($|/)'
#  - '/\.git($|/)'

# A file glob (or list of file globs) that will be matched against the file
# path before syncing the modules and states to the minions. This is similar
# to file_ignore_regex above, but works on globs instead of regex. By default
# nothing is ignored.
#
# file_ignore_glob:
#  - '*.pyc'
#  - '*/somefolder/*.bak'
#  - '*.swp'

# File Server Backend
# Salt supports a modular fileserver backend system, this system allows
# the salt master to link directly to third party systems to gather and
# manage the files available to minions. Multiple backends can be
# configured and will be searched for the requested file in the order in which
# they are defined here. The default setting only enables the standard backend
# "roots" which uses the "file_roots" option.
#
#fileserver_backend:
#  - roots
#
# To use multiple backends list them in the order they are searched:
#
#fileserver_backend:
#  - git
#  - roots

# Git fileserver backend configuration
# When using the git fileserver backend at least one git remote needs to be
# defined. The user running the salt master will need read access to the repo.
#
#gitfs_remotes:
#  - git://github.com/saltstack/salt-states.git
#  - file:///var/git/saltmaster
#
# The repos will be searched in order to find the file requested by a client
# and the first repo to have the file will return it.
# When using the git backend branches and tags are translated into salt
# environments.
# Note:  file:// repos will be treated as a remote, so refs you want used must
# exist in that repo as *local* refs.
#     
# The gitfs_root option gives the ability to serve files from a subdirectory
# within the repository. The path is defined relative to the root of the
# repository and defaults to the repository root.
#gitfs_root: somefolder/otherfolder


#####         Pillar settings        #####
##########################################
# Salt Pillars allow for the building of global data that can be made selectively
# available to different minions based on minion grain filtering. The Salt
# Pillar is laid out in the same fashion as the file server, with environments,
# a top file and sls files. However, pillar data does not need to be in the
# highstate format, and is generally just key/value pairs.

#pillar_roots:
#  base:
#    - /srv/pillar

#ext_pillar:
#  - hiera: /etc/hiera.yaml
#  - cmd_yaml: cat /etc/salt/yaml

# The pillar_opts option adds the master configuration file data to a dict in
# the pillar called "master". This is used to set simple configurations in the
# master config file that can then be used on minions.
#pillar_opts: True


#####          Syndic settings       #####
##########################################
# The Salt syndic is used to pass commands through a master from a higher
# master. Using the syndic is simple, if this is a master that will have
# syndic servers(s) below it set the "order_masters" setting to True, if this
# is a master that will be running a syndic daemon for passthrough the
# "syndic_master" setting needs to be set to the location of the master server
# to receive commands from.

# Set the order_masters setting to True if this master will command lower
# masters' syndic interfaces.
#order_masters: False

# If this master will be running a salt syndic daemon, syndic_master tells
# this master where to receive commands from.
#syndic_master: masterofmaster

# This is the 'ret_port' of the MasterOfMaster
#syndic_master_port: 4506

# PID file of the syndic daemon
#syndic_pidfile: /var/run/salt-syndic.pid

# LOG file of the syndic daemon
#syndic_log_file: syndic.log

#####      Peer Publish settings     #####
##########################################
# Salt minions can send commands to other minions, but only if the minion is
# allowed to. By default "Peer Publication" is disabled, and when enabled it
# is enabled for specific minions and specific commands. This allows secure
# compartmentalization of commands based on individual minions.

# The configuration uses regular expressions to match minions and then a list
# of regular expressions to match functions. The following will allow the
# minion authenticated as foo.example.com to execute functions from the test
# and pkg modules.
#
#peer:
#  foo.example.com:
#    - test.*
#    - pkg.*
#
# This will allow all minions to execute all commands:
#
#peer:
#  .*:
#    - .*
#
# This is not recommended, since it would allow anyone who gets root on any
# single minion to instantly have root on all of the minions!

# Minions can also be allowed to execute runners from the salt master.
# Since executing a runner from the minion could be considered a security risk,
# it needs to be enabled. This setting functions just like the peer setting
# except that it opens up runners instead of module functions.
#
# All peer runner support is turned off by default and must be enabled before
# using. This will enable all peer runners for all minions:
#
#peer_run:
#  .*:
#    - .*
#
# To enable just the manage.up runner for the minion foo.example.com:
#
#peer_run:
#  foo.example.com:
#    - manage.up


#####         Logging settings       #####
##########################################
# The location of the master log file
# The master log can be sent to a regular file, local path name, or network
# location. Remote logging works best when configured to use rsyslogd(8) (e.g.:
# ``file:///dev/log``), with rsyslogd(8) configured for network logging. The URI
# format is: <file|udp|tcp>://<host|socketpath>:<port-if-required>/<log-facility>
#log_file: /var/log/salt/master
#log_file: file:///dev/log
#log_file: udp://loghost:10514

#log_file: /var/log/salt/master
#key_logfile: /var/log/salt/key

# The level of messages to send to the console.
# One of 'garbage', 'trace', 'debug', info', 'warning', 'error', 'critical'.
#log_level: warning

# The level of messages to send to the log file.
# One of 'garbage', 'trace', 'debug', info', 'warning', 'error', 'critical'.
#log_level_logfile: warning

# The date and time format used in log messages. Allowed date/time formating
# can be seen here: http://docs.python.org/library/time.html#time.strftime
#log_datefmt: '%H:%M:%S'
#log_datefmt_logfile: '%Y-%m-%d %H:%M:%S'

# The format of the console logging messages. Allowed formatting options can
# be seen here: http://docs.python.org/library/logging.html#logrecord-attributes
#log_fmt_console: '[%(levelname)-8s] %(message)s'
#log_fmt_logfile: '%(asctime)s,%(msecs)03.0f [%(name)-17s][%(levelname)-8s] %(message)s'

# This can be used to control logging levels more specificically.  This
# example sets the main salt library at the 'warning' level, but sets
# 'salt.modules' to log at the 'debug' level:
#   log_granular_levels:
#     'salt': 'warning',
#     'salt.modules': 'debug'
#
#log_granular_levels: {}


#####         Node Groups           #####
##########################################
# Node groups allow for logical groupings of minion nodes.
# A group consists of a group name and a compound target.
#
#nodegroups:
#  group1: 'L@foo.domain.com,bar.domain.com,baz.domain.com and bl*.domain.com'
#  group2: 'G@os:Debian and foo.domain.com'


#####     Range Cluster settings     #####
##########################################
# The range server (and optional port) that serves your cluster information
# https://github.com/grierj/range/wiki/Introduction-to-Range-with-YAML-files
#
#range_server: range:80


#####     Windows Software Repo settings #####
##############################################
# Location of the repo on the master
#win_repo: '/srv/salt/win/repo'

# Location of the master's repo cache file
#win_repo_mastercachefile: '/srv/salt/win/repo/winrepo.p'

# List of git repositories to include with the local repo
#win_gitrepos:
#  - 'https://github.com/saltstack/salt-winrepo.git'

Example minion configuration file ¶

##### Primary configuration settings #####
##########################################

# Per default the minion will automatically include all config files
# from minion.d/*.conf (minion.d is a directory in the same directory
# as the main minion config file).
#default_include: minion.d/*.conf

# Set the location of the salt master server, if the master server cannot be
# resolved, then the minion will fail to start.
#master: salt

# Set whether the minion should connect to the master via IPv6
#ipv6: False

# Set the number of seconds to wait before attempting to resolve
# the master hostname if name resolution fails. Defaults to 30 seconds.
# Set to zero if the minion should shutdown and not retry.
# retry_dns: 30

# Set the port used by the master reply and authentication server
#master_port: 4506

# The user to run salt
#user: root

# Specify the location of the daemon process ID file
#pidfile: /var/run/salt-minion.pid

# The root directory prepended to these options: pki_dir, cachedir, log_file,
# sock_dir, pidfile.
#root_dir: /

# The directory to store the pki information in
#pki_dir: /etc/salt/pki/minion

# Explicitly declare the id for this minion to use, if left commented the id
# will be the hostname as returned by the python call: socket.getfqdn()
# Since salt uses detached ids it is possible to run multiple minions on the
# same machine but with different ids, this can be useful for salt compute
# clusters.
#id:

# Append a domain to a hostname in the event that it does not exist.  This is
# useful for systems where socket.getfqdn() does not actually result in a
# FQDN (for instance, Solaris).
#append_domain:

# Custom static grains for this minion can be specified here and used in SLS
# files just like all other grains. This example sets 4 custom grains, with
# the 'roles' grain having two values that can be matched against:
#grains:
#  roles:
#    - webserver
#    - memcache
#  deployment: datacenter4
#  cabinet: 13
#  cab_u: 14-15

# Where cache data goes
#cachedir: /var/cache/salt/minion

# Verify and set permissions on configuration directories at startup
#verify_env: True

# The minion can locally cache the return data from jobs sent to it, this
# can be a good way to keep track of jobs the minion has executed
# (on the minion side). By default this feature is disabled, to enable
# set cache_jobs to True
#cache_jobs: False

# set the directory used to hold unix sockets
#sock_dir: /var/run/salt/minion

# Set the default outputter used by the salt-call command. The default is
# "nested"
#output: nested
#
# By default output is colored, to disable colored output set the color value
# to False
#color: True

# Backup files that are replaced by file.managed and file.recurse under
# 'cachedir'/file_backups relative to their original location and appended
# with a timestamp. The only valid setting is "minion". Disabled by default.
#
# Alternatively this can be specified for each file in state files:
#
# /etc/ssh/sshd_config:
#   file.managed:
#     - source: salt://ssh/sshd_config
#     - backup: minion
#
#backup_mode: minion

# When waiting for a master to accept the minion's public key, salt will
# continuously attempt to reconnect until successful. This is the time, in
# seconds, between those reconnection attempts.
#acceptance_wait_time: 10

# If this is nonzero, the time between reconnection attempts will increase by
# acceptance_wait_time seconds per iteration, up to this maximum. If this is
# set to zero, the time between reconnection attempts will stay constant.
#acceptance_wait_time_max: 0

# When the master key changes, the minion will try to re-auth itself to receive
# the new master key. In larger environments this can cause a SYN flood on the
# master because all minions try to re-auth immediately. To prevent this and
# have a minion wait for a random amount of time, use this optional parameter.
# The wait-time will be a random number of seconds between
# 0 and the defined value.
#random_reauth_delay: 60


# If you don't have any problems with syn-floods, dont bother with the
# three recon_* settings described below, just leave the defaults!
#
# The ZeroMQ pull-socket that binds to the masters publishing interface tries
# to reconnect immediately, if the socket is disconnected (for example if
# the master processes are restarted). In large setups this will have all
# minions reconnect immediately which might flood the master (the ZeroMQ-default
# is usually a 100ms delay). To prevent this, these three recon_* settings 
# can be used.
#
# recon_default: the interval in milliseconds that the socket should wait before 
#                trying to reconnect to the master (100ms = 1 second)
#
# recon_max: the maximum time a socket should wait. each interval the time to wait
#            is calculated by doubling the previous time. if recon_max is reached,
#            it starts again at recon_default. Short example:
#
#            reconnect 1: the socket will wait 'recon_default' milliseconds
#            reconnect 2: 'recon_default' * 2
#            reconnect 3: ('recon_default' * 2) * 2
#            reconnect 4: value from previous interval * 2
#            reconnect 5: value from previous interval * 2
#            reconnect x: if value >= recon_max, it starts again with recon_default
#
# recon_randomize: generate a random wait time on minion start. The wait time will 
#                  be a random value between recon_default and recon_default + 
#                  recon_max. Having all minions reconnect with the same recon_default 
#                  and recon_max value kind of defeats the purpose of being able to 
#                  change these settings. If all minions have the same values and your 
#                  setup is quite large (several thousand minions), they will still 
#                  flood the master. The desired behaviour is to have timeframe within
#                  all minions try to reconnect. 

# Example on how to use these settings:
# The goal: have all minions reconnect within a 60 second timeframe on a disconnect
#
# The settings:
#recon_default: 1000
#recon_max: 59000
#recon_randomize: True
#
# Each minion will have a randomized reconnect value between 'recon_default'
# and 'recon_default + recon_max', which in this example means between 1000ms
# 60000ms (or between 1 and 60 seconds). The generated random-value will be 
# doubled after each attempt to reconnect. Lets say the generated random 
# value is 11 seconds (or 11000ms). 
#
# reconnect 1: wait 11 seconds
# reconnect 2: wait 22 seconds
# reconnect 3: wait 33 seconds
# reconnect 4: wait 44 seconds
# reconnect 5: wait 55 seconds
# reconnect 6: wait time is bigger than 60 seconds (recon_default + recon_max)
# reconnect 7: wait 11 seconds
# reconnect 8: wait 22 seconds
# reconnect 9: wait 33 seconds
# reconnect x: etc.
#
# In a setup with ~6000 thousand hosts these settings would average the reconnects
# to about 100 per second and all hosts would be reconnected within 60 seconds.
#recon_default: 100
#recon_max: 5000
#recon_randomize: False

# The loop_interval sets how long in seconds the minion will wait between
# evaluating the scheduler and running cleanup tasks. This defaults to a
# sane 60 seconds, but if the minion scheduler needs to be evaluated more
# often lower this value
#loop_interval: 60

# When healing, a dns_check is run. This is to make sure that the originally
# resolved dns has not changed. If this is something that does not happen in
# your environment, set this value to False.
#dns_check: True

# Windows platforms lack posix IPC and must rely on slower TCP based inter-
# process communications. Set ipc_mode to 'tcp' on such systems
#ipc_mode: ipc
#
# Overwrite the default tcp ports used by the minion when in tcp mode
#tcp_pub_port: 4510
#tcp_pull_port: 4511

# The minion can include configuration from other files. To enable this,
# pass a list of paths to this option. The paths can be either relative or
# absolute; if relative, they are considered to be relative to the directory
# the main minion configuration file lives in (this file). Paths can make use
# of shell-style globbing. If no files are matched by a path passed to this
# option then the minion will log a warning message.
#
#
# Include a config file from some other path:
# include: /etc/salt/extra_config
#
# Include config from several files and directories:
#include:
#  - /etc/salt/extra_config
#  - /etc/roles/webserver

#####   Minion module management     #####
##########################################
# Disable specific modules. This allows the admin to limit the level of
# access the master has to the minion
#disable_modules: [cmd,test]
#disable_returners: []
#
# Modules can be loaded from arbitrary paths. This enables the easy deployment
# of third party modules. Modules for returners and minions can be loaded.
# Specify a list of extra directories to search for minion modules and
# returners. These paths must be fully qualified!
#module_dirs: []
#returner_dirs: []
#states_dirs: []
#render_dirs: []
#
# A module provider can be statically overwritten or extended for the minion
# via the providers option, in this case the default module will be
# overwritten by the specified module. In this example the pkg module will
# be provided by the yumpkg5 module instead of the system default.
#
#providers:
#  pkg: yumpkg5
#
# Enable Cython modules searching and loading. (Default: False)
#cython_enable: False
#

#####    State Management Settings    #####
###########################################
# The state management system executes all of the state templates on the minion
# to enable more granular control of system state management. The type of
# template and serialization used for state management needs to be configured
# on the minion, the default renderer is yaml_jinja. This is a yaml file
# rendered from a jinja template, the available options are:
# yaml_jinja
# yaml_mako
# yaml_wempy
# json_jinja
# json_mako
# json_wempy
#
#renderer: yaml_jinja
#
# The failhard option tells the minions to stop immediately after the first
# failure detected in the state execution, defaults to False
#failhard: False
#
# autoload_dynamic_modules Turns on automatic loading of modules found in the
# environments on the master. This is turned on by default, to turn of
# autoloading modules when states run set this value to False
#autoload_dynamic_modules: True
#
# clean_dynamic_modules keeps the dynamic modules on the minion in sync with
# the dynamic modules on the master, this means that if a dynamic module is
# not on the master it will be deleted from the minion. By default this is
# enabled and can be disabled by changing this value to False
#clean_dynamic_modules: True
#
# Normally the minion is not isolated to any single environment on the master
# when running states, but the environment can be isolated on the minion side
# by statically setting it. Remember that the recommended way to manage
# environments is to isolate via the top file.
#environment: None
#
# If using the local file directory, then the state top file name needs to be
# defined, by default this is top.sls.
#state_top: top.sls
#
# Run states when the minion daemon starts. To enable, set startup_states to:
# 'highstate' -- Execute state.highstate
# 'sls' -- Read in the sls_list option and execute the named sls files
# 'top' -- Read top_file option and execute based on that file on the Master
#startup_states: ''
#
# list of states to run when the minion starts up if startup_states is 'sls'
#sls_list:
#  - edit.vim
#  - hyper
#
# top file to execute if startup_states is 'top'
#top_file: ''

#####     File Directory Settings    #####
##########################################
# The Salt Minion can redirect all file server operations to a local directory,
# this allows for the same state tree that is on the master to be used if
# copied completely onto the minion. This is a literal copy of the settings on
# the master but used to reference a local directory on the minion.

# Set the file client. The client defaults to looking on the master server for
# files, but can be directed to look at the local file directory setting
# defined below by setting it to local.
#file_client: remote

# The file directory works on environments passed to the minion, each environment
# can have multiple root directories, the subdirectories in the multiple file
# roots cannot match, otherwise the downloaded files will not be able to be
# reliably ensured. A base environment is required to house the top file.
# Example:
# file_roots:
#   base:
#     - /srv/salt/
#   dev:
#     - /srv/salt/dev/services
#     - /srv/salt/dev/states
#   prod:
#     - /srv/salt/prod/services
#     - /srv/salt/prod/states
#
#file_roots:
#  base:
#    - /srv/salt

# The hash_type is the hash to use when discovering the hash of a file in
# the local fileserver. The default is md5, but sha1, sha224, sha256, sha384
# and sha512 are also supported.
#hash_type: md5

# The Salt pillar is searched for locally if file_client is set to local. If
# this is the case, and pillar data is defined, then the pillar_roots need to
# also be configured on the minion:
#pillar_roots:
#  base:
#    - /srv/pillar

######        Security settings       #####
###########################################
# Enable "open mode", this mode still maintains encryption, but turns off
# authentication, this is only intended for highly secure environments or for
# the situation where your keys end up in a bad state. If you run in open mode
# you do so at your own risk!
#open_mode: False

# Enable permissive access to the salt keys.  This allows you to run the
# master or minion as root, but have a non-root group be given access to
# your pki_dir.  To make the access explicit, root must belong to the group
# you've given access to. This is potentially quite insecure.
#permissive_pki_access: False

# The state_verbose and state_output settings can be used to change the way
# state system data is printed to the display. By default all data is printed.
# The state_verbose setting can be set to True or False, when set to False
# all data that has a result of True and no changes will be suppressed.
#state_verbose: True
#
# The state_output setting changes if the output is the full multi line
# output for each changed state if set to 'full', but if set to 'terse'
# the output will be shortened to a single line.
#state_output: full
#
# Fingerprint of the master public key to double verify the master is valid,
# the master fingerprint can be found by running "salt-key -F master" on the
# salt master.
#master_finger: ''

######         Thread settings        #####
###########################################
# Disable multiprocessing support, by default when a minion receives a
# publication a new process is spawned and the command is executed therein.
#multiprocessing: True

#####         Logging settings       #####
##########################################
# The location of the minion log file
# The minion log can be sent to a regular file, local path name, or network
# location. Remote logging works best when configured to use rsyslogd(8) (e.g.:
# ``file:///dev/log``), with rsyslogd(8) configured for network logging. The URI
# format is: <file|udp|tcp>://<host|socketpath>:<port-if-required>/<log-facility>
#log_file: /var/log/salt/minion
#log_file: file:///dev/log
#log_file: udp://loghost:10514
#
#log_file: /var/log/salt/minion
#key_logfile: /var/log/salt/key
#
# The level of messages to send to the console.
# One of 'garbage', 'trace', 'debug', info', 'warning', 'error', 'critical'.
# Default: 'warning'
#log_level: warning
#
# The level of messages to send to the log file.
# One of 'garbage', 'trace', 'debug', info', 'warning', 'error', 'critical'.
# Default: 'warning'
#log_level_logfile:

# The date and time format used in log messages. Allowed date/time formating
# can be seen here: http://docs.python.org/library/time.html#time.strftime
#log_datefmt: '%H:%M:%S'
#log_datefmt_logfile: '%Y-%m-%d %H:%M:%S'
#
# The format of the console logging messages. Allowed formatting options can
# be seen here: http://docs.python.org/library/logging.html#logrecord-attributes
#log_fmt_console: '[%(levelname)-8s] %(message)s'
#log_fmt_logfile: '%(asctime)s,%(msecs)03.0f [%(name)-17s][%(levelname)-8s] %(message)s'
#
# This can be used to control logging levels more specificically.  This
# example sets the main salt library at the 'warning' level, but sets
# 'salt.modules' to log at the 'debug' level:
#   log_granular_levels:
#     'salt': 'warning',
#     'salt.modules': 'debug'
#
#log_granular_levels: {}

######      Module configuration      #####
###########################################
# Salt allows for modules to be passed arbitrary configuration data, any data
# passed here in valid yaml format will be passed on to the salt minion modules
# for use. It is STRONGLY recommended that a naming convention be used in which
# the module name is followed by a . and then the value. Also, all top level
# data must be applied via the yaml dict construct, some examples:
#
# You can specify that all modules should run in test mode:
#test: True
#
# A simple value for the test module:
#test.foo: foo
#
# A list for the test module:
#test.bar: [baz,quo]
#
# A dict for the test module:
#test.baz: {spam: sausage, cheese: bread}


######      Update settings          ######
###########################################
# Using the features in Esky, a salt minion can both run as a frozen app and
# be updated on the fly. These options control how the update process
# (saltutil.update()) behaves.
#
# The url for finding and downloading updates. Disabled by default.
#update_url: False
#
# The list of services to restart after a successful update. Empty by default.
#update_restart_services: []


######      Keepalive settings        ######
############################################
# ZeroMQ now includes support for configuring SO_KEEPALIVE if supported by
# the OS. If connections between the minion and the master pass through
# a state tracking device such as a firewall or VPN gateway, there is
# the risk that it could tear down the connection the master and minion
# without informing either party that their connection has been taken away.
# Enabling TCP Keepalives prevents this from happening.
#
# Overall state of TCP Keepalives, enable (1 or True), disable (0 or False)
# or leave to the OS defaults (-1), on Linux, typically disabled. Default True, enabled.
#tcp_keepalive: True
#
# How long before the first keepalive should be sent in seconds. Default 300
# to send the first keepalive after 5 minutes, OS default (-1) is typically 7200 seconds
# on Linux see /proc/sys/net/ipv4/tcp_keepalive_time.
#tcp_keepalive_idle: 300
#
# How many lost probes are needed to consider the connection lost. Default -1
# to use OS defaults, typically 9 on Linux, see /proc/sys/net/ipv4/tcp_keepalive_probes.
#tcp_keepalive_cnt: -1
#
# How often, in seconds, to send keepalives after the first one. Default -1 to
# use OS defaults, typically 75 seconds on Linux, see
# /proc/sys/net/ipv4/tcp_keepalive_intvl.
#tcp_keepalive_intvl: -1


######      Windows Software settings ######
############################################
# Location of the repository cache file on the master
#win_repo_cachefile: 'salt://win/repo/winrepo.p'

Configuring the Salt Master¶

The Salt system is amazingly simple and easy to configure, the two components of the Salt system each have a respective configuration file. The salt-master is configured via the master configuration file, and the salt-minion is configured via the minion configuration file.

See also

example master configuration file

The configuration file for the salt-master is located at /etc/salt/master. The available options are as follows:

Primary Master Configuration¶

`interface`¶

Default: 0.0.0.0 (all interfaces)

The local interface to bind to.

interface: 192.168.0.1

`publish_port`¶

Default: 4505

The network port to set up the publication interface

publish_port: 4505

`user`¶

Default: root

The user to run the Salt processes

user: root

`max_open_files`¶

Default: max_open_files

Each minion connecting to the master uses AT LEAST one file descriptor, the master subscription connection. If enough minions connect you might start seeing on the console(and then salt-master crashes):

Too many open files (tcp_listener.cpp:335)
Aborted (core dumped)

By default this value will be the one of ulimit -Hn, i.e., the hard limit for max open files.

If you wish to set a different value than the default one, uncomment and configure this setting. Remember that this value CANNOT be higher than the hard limit. Raising the hard limit depends on your OS and/or distribution, a good way to find the limit is to search the internet for(for example):

raise max open files hard limit debian

max_open_files: 100000

`worker_threads`¶

Default: 5

The number of threads to start for receiving commands and replies from minions. If minions are stalling on replies because you have many minions, raise the worker_threads value.

Worker threads should not be put below 3 when using the peer system, but can drop down to 1 worker otherwise.

worker_threads: 5

`ret_port`¶

Default: 4506

The port used by the return server, this is the server used by Salt to receive execution returns and command executions.

ret_port: 4506

`pidfile`¶

Default: /var/run/salt-master.pid

Specify the location of the master pidfile

pidfile: /var/run/salt-master.pid

`root_dir`¶

Default: /

The system root directory to operate from, change this to make Salt run from an alternative root

root_dir: /

`pki_dir`¶

Default: /etc/salt/pki

The directory to store the pki authentication keys.

pki_dir: /etc/salt/pki

`cachedir`¶

Default: /var/cache/salt

The location used to store cache information, particularly the job information for executed salt commands.

cachedir: /var/cache/salt

`keep_jobs`¶

Default: 24

Set the number of hours to keep old job information

`job_cache`¶

Default: True

The master maintains a job cache, while this is a great addition it can be a burden on the master for larger deployments (over 5000 minions). Disabling the job cache will make previously executed jobs unavailable to the jobs system and is not generally recommended. Normally it is wise to make sure the master has access to a faster IO system or a tmpfs is mounted to the jobs dir

`ext_job_cache`¶

Default: ''

Used to specify a default returner for all minions, when this option is set the specified returner needs to be properly configured and the minions will always default to sending returns to this returner. This will also disable the local job cache on the master

ext_job_cache: redis

`minion_data_cache`¶

Default: True

The minion data cache is a cache of information about the minions stored on the master, this information is primarily the pillar and grains data. The data is cached in the Master cachedir under the name of the minion and used to pre determine what minions are expected to reply from executions.

minion_cache_dir: True

`enforce_mine_cache`¶

Default: False

By-default when disabling the minion_data_cache mine will stop working since it is based on cached data, by enabling this option we explicitly enabling only the cache for the mine system.

enforce_mine_cache: False

`sock_dir`¶

Default:: /tmp/salt-unix

Set the location to use for creating Unix sockets for master process communication

Master Security Settings¶

`open_mode`¶

Default: False

Open mode is a dangerous security feature. One problem encountered with pki authentication systems is that keys can become "mixed up" and authentication begins to fail. Open mode turns off authentication and tells the master to accept all authentication. This will clean up the pki keys received from the minions. Open mode should not be turned on for general use. Open mode should only be used for a short period of time to clean up pki keys. To turn on open mode set this value to True.

open_mode: False

`auto_accept`¶

Default: False

Enable auto_accept. This setting will automatically accept all incoming public keys from the minions

auto_accept: False

`autosign_file`¶

Default not defined

If the autosign_file is specified incoming keys specified in the autosign_file will be automatically accepted. Matches will be searched for first by string comparison, then by globbing, then by full-string regex matching. This is insecure!

`client_acl`¶

Default: {}

Enable user accounts on the master to execute specific modules. These modules can be expressed as regular expressions

client_acl:
  fred:
    - test.ping
    - pkg.*

`client_acl_blacklist`¶

Default: {}

Blacklist users or modules

This example would blacklist all non sudo users, including root from running any commands. It would also blacklist any use of the "cmd" module.

This is completely disabled by default.

client_acl_blacklist:
  users:
    - root
    - '^(?!sudo_).*$'   #  all non sudo users
  modules:
    - cmd

`external_auth`¶

Default: {}

The external auth system uses the Salt auth modules to authenticate and validate users to access areas of the Salt system.

external_auth:
  pam:
    fred:
      - test.*

`token_expire`¶

Default: 43200

Time (in seconds) for a newly generated token to live. Default: 12 hours

token_expire: 43200

`file_recv`¶

Default: False

Allow minions to push files to the master. This is disabled by default, for security purposes.

file_recv: False

Master Module Management¶

`runner_dirs`¶

Default: []

Set additional directories to search for runner modules

`cython_enable`¶

Default: False

Set to true to enable cython modules (.pyx files) to be compiled on the fly on the Salt master

cython_enable: False

Master State System Settings¶

`state_verbose`¶

Default: False

state_verbose allows for the data returned from the minion to be more verbose. Normally only states that fail or states that have changes are returned, but setting state_verbose to True will return all states that were checked

state_verbose: True

`state_output`¶

Default: full

The state_output setting changes if the output is the full multi line output for each changed state if set to 'full', but if set to 'terse' the output will be shortened to a single line. If set to 'mixed', the output will be terse unless a state failed, in which case that output will be full. If set to 'changes', the output will be full unless the state didn't change.

state_output: full

`state_top`¶

Default: top.sls

The state system uses a "top" file to tell the minions what environment to use and what modules to use. The state_top file is defined relative to the root of the base environment

state_top: top.sls

`external_nodes`¶

Default: None

The external_nodes option allows Salt to gather data that would normally be placed in a top file from and external node controller. The external_nodes option is the executable that will return the ENC data. Remember that Salt will look for external nodes AND top files and combine the results if both are enabled and available!

external_nodes: cobbler-ext-nodes

`renderer`¶

Default: yaml_jinja

The renderer to use on the minions to render the state data

renderer: yaml_jinja

`failhard`¶

Default:: False

Set the global failhard flag, this informs all states to stop running states at the moment a single state fails

failhard: False

`test`¶

Default:: False

Set all state calls to only test if they are going to actually make changes or just post what changes are going to be made

test: False

Master File Server Settings¶

`file_roots`¶

Default:

base:
  - /srv/salt

Salt runs a lightweight file server written in ZeroMQ to deliver files to minions. This file server is built into the master daemon and does not require a dedicated port.

The file server works on environments passed to the master. Each environment can have multiple root directories. The subdirectories in the multiple file roots cannot match, otherwise the downloaded files will not be able to be reliably ensured. A base environment is required to house the top file. Example:

file_roots:
  base:
    - /srv/salt
  dev:
    - /srv/salt/dev/services
    - /srv/salt/dev/states
  prod:
    - /srv/salt/prod/services
    - /srv/salt/prod/states

`hash_type`¶

Default: md5

The hash_type is the hash to use when discovering the hash of a file on the master server. The default is md5, but sha1, sha224, sha256, sha384 and sha512 are also supported.

hash_type: md5

`file_buffer_size`¶

Default: 1048576

The buffer size in the file server in bytes

file_buffer_size: 1048576

Pillar Configuration¶

`pillar_roots`¶

Default:

base:
  - /srv/pillar

Set the environments and directories used to hold pillar sls data. This configuration is the same as file_roots:

pillar_roots:
  base:
    - /srv/pillar
  dev:
    - /srv/pillar/dev
  prod:
    - /srv/pillar/prod

`ext_pillar`¶

The ext_pillar option allows for any number of external pillar interfaces to be called when populating pillar data. The configuration is based on ext_pillar functions. The available ext_pillar functions are: hiera, cmd_yaml. By default the ext_pillar interface is not configured to run.

Default:: None

ext_pillar:
  - hiera: /etc/hiera.yaml
  - cmd_yaml: cat /etc/salt/yaml
  - reclass:
      inventory_base_uri: /etc/reclass

There are additional details at Pillars

Syndic Server Settings¶

A Salt syndic is a Salt master used to pass commands from a higher Salt master to minions below the syndic. Using the syndic is simple. If this is a master that will have syndic servers(s) below it, set the "order_masters" setting to True. If this is a master that will be running a syndic daemon for passthrough the "syndic_master" setting needs to be set to the location of the master server

Do not not forget that in other word it means that it shares with the local minion it's ID and PKI_DIR.

`order_masters`¶

Default: False

Extra data needs to be sent with publications if the master is controlling a lower level master via a syndic minion. If this is the case the order_masters value must be set to True

order_masters: False

`syndic_master`¶

Default: None

If this master will be running a salt-syndic to connect to a higher level master, specify the higher level master with this configuration value

syndic_master: masterofmasters

`syndic_master_port`¶

Default: 4506

If this master will be running a salt-syndic to connect to a higher level master, specify the higher level master port with this configuration value

syndic_master_port: 4506

`syndic_log_file`¶

Default: syndic.log

If this master will be running a salt-syndic to connect to a higher level master, specify the log_file of the syndic daemon.

syndic_log_file: salt-syndic.log

`syndic_pidfile`¶

Default: salt-syndic.pid

If this master will be running a salt-syndic to connect to a higher level master, specify the pidfile of the syndic daemon.

syndic_pidfile: syndic.pid

Peer Publish Settings¶

Salt minions can send commands to other minions, but only if the minion is allowed to. By default "Peer Publication" is disabled, and when enabled it is enabled for specific minions and specific commands. This allows secure compartmentalization of commands based on individual minions.

`peer`¶

Default: {}

The configuration uses regular expressions to match minions and then a list of regular expressions to match functions. The following will allow the minion authenticated as foo.example.com to execute functions from the test and pkg modules

peer:
  foo.example.com:
      - test.*
      - pkg.*

This will allow all minions to execute all commands:

peer:
  .*:
      - .*

This is not recommended, since it would allow anyone who gets root on any single minion to instantly have root on all of the minions!

`peer_run`¶

Default: {}

The peer_run option is used to open up runners on the master to access from the minions. The peer_run configuration matches the format of the peer configuration.

The following example would allow foo.example.com to execute the manage.up runner:

peer_run:
  foo.example.com:
      - manage.up

Node Groups¶

Default: {}

Node groups allow for logical groupings of minion nodes. A group consists of a group name and a compound target.

nodegroups:
  group1: 'L@foo.domain.com,bar.domain.com,baz.domain.com or bl*.domain.com'
  group2: 'G@os:Debian and foo.domain.com'

Master Logging Settings¶

`log_file`¶

Default: /var/log/salt/master

The master log can be sent to a regular file, local path name, or network location. See also log_file.

Examples:

log_file: /var/log/salt/master

log_file: file:///dev/log

log_file: udp://loghost:10514

`log_level`¶

Default: warning

The level of messages to send to the console. See also log_level.

log_level: warning

`log_level_logfile`¶

Default: warning

The level of messages to send to the log file. See also log_level_logfile.

log_level_logfile: warning

`log_datefmt`¶

Default: %H:%M:%S

The date and time format used in console log messages. See also log_datefmt.

log_datefmt: '%H:%M:%S'

`log_datefmt_logfile`¶

Default: %Y-%m-%d %H:%M:%S

The date and time format used in log file messages. See also log_datefmt_logfile.

log_datefmt_logfile: '%Y-%m-%d %H:%M:%S'

`log_fmt_console`¶

Default: [%(levelname)-8s] %(message)s

The format of the console logging messages. See also log_fmt_console.

log_fmt_console: '[%(levelname)-8s] %(message)s'

`log_fmt_logfile`¶

Default: %(asctime)s,%(msecs)03.0f [%(name)-17s][%(levelname)-8s] %(message)s

The format of the log file logging messages. See also log_fmt_logfile.

log_fmt_logfile: '%(asctime)s,%(msecs)03.0f [%(name)-17s][%(levelname)-8s] %(message)s'

`log_granular_levels`¶

Default: {}

This can be used to control logging levels more specifically. See also log_granular_levels.

Include Configuration¶

`default_include`¶

Default: master.d/*.conf

The master can include configuration from other files. Per default the master will automatically include all config files from master.d/*.conf where master.d is relative to the directory of the master configuration file.

`include`¶

Default: not defined

The master can include configuration from other files. To enable this, pass a list of paths to this option. The paths can be either relative or absolute; if relative, they are considered to be relative to the directory the main minion configuration file lives in. Paths can make use of shell-style globbing. If no files are matched by a path passed to this option then the master will log a warning message.

# Include files from a master.d directory in the same
# directory as the master config file
include: master.d/*

# Include a single extra file into the configuration
include: /etc/roles/webserver

# Include several files and the master.d directory
include:
  - extra_config
  - master.d/*
  - /etc/roles/webserver

Configuring the Salt Minion¶

The Salt system is amazingly simple and easy to configure, the two components of the Salt system each have a respective configuration file. The salt-master is configured via the master configuration file, and the salt-minion is configured via the minion configuration file.

See also

example minion configuration file

The Salt Minion configuration is very simple, typically the only value that needs to be set is the master value so the minion can find its master.

Minion Primary Configuration¶

`master`¶

Default: salt

The hostname or ipv4 of the master.

master: salt

`master_port`¶

Default: 4506

The port of the master ret server, this needs to coincide with the ret_port option on the Salt master.

master_port: 4506

`user`¶

Default: root

The user to run the Salt processes

user: root

`pidfile`¶

Default: /var/run/salt-minion.pid

The location of the daemon's process ID file

pidfile: /var/run/salt-minion.pid

`root_dir`¶

Default: /

This directory is prepended to the following options: pki_dir, cachedir, log_file, sock_dir, and pidfile.

root_dir: /

`pki_dir`¶

Default: /etc/salt/pki

The directory used to store the minion's public and private keys.

pki_dir: /etc/salt/pki

`id`¶

Default: the system's hostname

See also

Salt Walkthrough

The Setting up a Salt Minion section contains detailed information on how the hostname is determined.

Explicitly declare the id for this minion to use. Since Salt uses detached ids it is possible to run multiple minions on the same machine but with different ids. This can be useful for Salt compute clusters.

id: foo.bar.com

`append_domain`¶

Default: None

Append a domain to a hostname in the event that it does not exist. This is useful for systems where socket.getfqdn() does not actually result in a FQDN (for instance, Solaris).

append_domain: foo.org

`cachedir`¶

Default: /var/cache/salt

The location for minion cache data.

cachedir: /var/cache/salt

`verify_env`¶

Default: True

Verify and set permissions on configuration directories at startup.

verify_env: True

`cache_jobs`¶

Default: False

The minion can locally cache the return data from jobs sent to it, this can be a good way to keep track of the minion side of the jobs the minion has executed. By default this feature is disabled, to enable set cache_jobs to True.

cache_jobs: False

`sock_dir`¶

Default: /var/run/salt/minion

The directory where Unix sockets will be kept.

sock_dir: /var/run/salt/minion

`backup_mode`¶

Default: []

Backup files replaced by file.managed and file.recurse under cachedir.

backup_mode: minion

`acceptance_wait_time`¶

Default: 10

The number of seconds to wait until attempting to re-authenticate with the master.

acceptance_wait_time: 10

`random_reauth_delay`¶

When the master key changes, the minion will try to re-auth itself to receive the new master key. In larger environments this can cause a syn-flood on the master because all minions try to re-auth immediately. To prevent this and have a minion wait for a random amount of time, use this optional parameter. The wait-time will be a random number of seconds between 0 and the defined value.

random_reauth_delay: 60

`acceptance_wait_time_max`¶

Default: None

The maximum number of seconds to wait until attempting to re-authenticate with the master. If set, the wait will increase by acceptance_wait_time seconds each iteration.

acceptance_wait_time_max: None

`dns_check`¶

Default: True

When healing, a dns_check is run. This is to make sure that the originally resolved dns has not changed. If this is something that does not happen in your environment, set this value to False.

dns_check: True

`ipc_mode`¶

Default: ipc

Windows platforms lack POSIX IPC and must rely on slower TCP based inter- process communications. Set ipc_mode to tcp on such systems.

ipc_mode: ipc

`tcp_pub_port`¶

Default: 4510

Publish port used when ipc_mode is set to tcp.

tcp_pub_port: 4510

`tcp_pull_port`¶

Default: 4511

Pull port used when ipc_mode is set to tcp.

tcp_pull_port: 4511

Minion Module Management¶

`disable_modules`¶

Default: [] (all modules are enabled by default)

The event may occur in which the administrator desires that a minion should not be able to execute a certain module. The sys module is built into the minion and cannot be disabled.

This setting can also tune the minion, as all modules are loaded into ram disabling modules will lover the minion's ram footprint.

disable_modules:
  - test
  - solr

`disable_returners`¶

Default: [] (all returners are enabled by default)

If certain returners should be disabled, this is the place

disable_returners:
  - mongo_return

`module_dirs`¶

Default: []

A list of extra directories to search for Salt modules

module_dirs:
  - /var/lib/salt/modules

`returner_dirs`¶

Default: []

A list of extra directories to search for Salt returners

returners_dirs:
  - /var/lib/salt/returners

`states_dirs`¶

Default: []

A list of extra directories to search for Salt states

states_dirs:
  - /var/lib/salt/states

`render_dirs`¶

Default: []

A list of extra directories to search for Salt renderers

render_dirs:
  - /var/lib/salt/renderers

`cython_enable`¶

Default: False

Set this value to true to enable auto-loading and compiling of .pyx modules, This setting requires that gcc and cython are installed on the minion

cython_enable: False

`providers`¶

Default: (empty)

A module provider can be statically overwritten or extended for the minion via the providers option. This can be done on an individual basis in an SLS file, or globally here in the minion config, like below.

providers:
  pkg: yumpkg5
  service: systemd

State Management Settings¶

`renderer`¶

Default: yaml_jinja

The default renderer used for local state executions

renderer: yaml_jinja

`state_verbose`¶

Default: False

state_verbose allows for the data returned from the minion to be more verbose. Normally only states that fail or states that have changes are returned, but setting state_verbose to True will return all states that were checked

state_verbose: True

`state_output`¶

Default: full

The state_output setting changes if the output is the full multi line output for each changed state if set to 'full', but if set to 'terse' the output will be shortened to a single line.

state_output: full

`autoload_dynamic_modules`¶

Default: True

autoload_dynamic_modules Turns on automatic loading of modules found in the environments on the master. This is turned on by default, to turn of auto-loading modules when states run set this value to False

autoload_dynamic_modules: True

Default: True

clean_dynamic_modules keeps the dynamic modules on the minion in sync with the dynamic modules on the master, this means that if a dynamic module is not on the master it will be deleted from the minion. By default this is enabled and can be disabled by changing this value to False

clean_dynamic_modules: True

`environment`¶

Default: None

Normally the minion is not isolated to any single environment on the master when running states, but the environment can be isolated on the minion side by statically setting it. Remember that the recommended way to manage environments is to isolate via the top file.

environment: None

File Directory Settings¶

`file_client`¶

Default: remote

The client defaults to looking on the master server for files, but can be directed to look on the minion by setting this parameter to local.

file_client: remote

`file_roots`¶

Default:

base:
  - /srv/salt

When using a local file_client, this parameter is used to setup the fileserver's environments. This parameter operates identically to the master config parameter of the same name.

file_roots:
  base:
    - /srv/salt
  dev:
    - /srv/salt/dev/services
    - /srv/salt/dev/states
  prod:
    - /srv/salt/prod/services
    - /srv/salt/prod/states

`hash_type`¶

Default: md5

The hash_type is the hash to use when discovering the hash of a file on the local fileserver. The default is md5, but sha1, sha224, sha256, sha384 and sha512 are also supported.

hash_type: md5

`pillar_roots`¶

Default:

base:
  - /srv/pillar

When using a local file_client, this parameter is used to setup the pillar environments.

pillar_roots:
  base:
    - /srv/pillar
  dev:
    - /srv/pillar/dev
  prod:
    - /srv/pillar/prod

Security Settings¶

`open_mode`¶

Default: False

Open mode can be used to clean out the PKI key received from the Salt master, turn on open mode, restart the minion, then turn off open mode and restart the minion to clean the keys.

open_mode: False

Thread Settings¶

Default: True

Disable multiprocessing support by default when a minion receives a publication a new process is spawned and the command is executed therein.

multiprocessing: True

Minion Logging Settings¶

`log_file`¶

Default: /var/log/salt/minion

The minion log can be sent to a regular file, local path name, or network location. See also log_file.

Examples:

log_file: /var/log/salt/minion

log_file: file:///dev/log

log_file: udp://loghost:10514

`log_level`¶

Default: warning

The level of messages to send to the console. See also log_level.

log_level: warning

`log_level_logfile`¶

Default: warning

The level of messages to send to the log file. See also log_level_logfile.

log_level_logfile: warning

`log_datefmt`¶

Default: %H:%M:%S

The date and time format used in console log messages. See also log_datefmt.

log_datefmt: '%H:%M:%S'

`log_datefmt_logfile`¶

Default: %Y-%m-%d %H:%M:%S

The date and time format used in log file messages. See also log_datefmt_logfile.

log_datefmt_logfile: '%Y-%m-%d %H:%M:%S'

`log_fmt_console`¶

Default: [%(levelname)-8s] %(message)s

The format of the console logging messages. See also log_fmt_console.

log_fmt_console: '[%(levelname)-8s] %(message)s'

`log_fmt_logfile`¶

Default: %(asctime)s,%(msecs)03.0f [%(name)-17s][%(levelname)-8s] %(message)s

The format of the log file logging messages. See also log_fmt_logfile.

log_fmt_logfile: '%(asctime)s,%(msecs)03.0f [%(name)-17s][%(levelname)-8s] %(message)s'

`log_granular_levels`¶

Default: {}

This can be used to control logging levels more specifically. See also log_granular_levels.

Include Configuration¶

`default_include`¶

Default: minion.d/*.conf

The minion can include configuration from other files. Per default the minion will automatically include all config files from minion.d/*.conf where minion.d is relative to the directory of the minion configuration file.

`include`¶

Default: not defined

The minion can include configuration from other files. To enable this, pass a list of paths to this option. The paths can be either relative or absolute; if relative, they are considered to be relative to the directory the main minion configuration file lives in. Paths can make use of shell-style globbing. If no files are matched by a path passed to this option then the minion will log a warning message.

# Include files from a minion.d directory in the same
# directory as the minion config file
include: minion.d/*

# Include a single extra file into the configuration
include: /etc/roles/webserver

# Include several files and the minion.d directory
include:
  - extra_config
  - minion.d/*
  - /etc/roles/webserver

Frozen Build Update Settings¶

These options control how salt.modules.saltutil.update() works with esky frozen apps. For more information look at https://github.com/cloudmatrix/esky/.

`update_url`¶

Default: False (Update feature is disabled)

The url to use when looking for application updates. Esky depends on directory listings to search for new versions. A webserver running on your Master is a good starting point for most setups.

update_url: 'http://salt.example.com/minion-updates'

`update_restart_services`¶

Default: [] (service restarting on update is disabled)

A list of services to restart when the minion software is updated. This would typically just be a list containing the minion's service name, but you may have other services that need to go with it.

update_restart_services: ['salt-minion']

Salt code and internals¶

Reference documentation on Salt's internal code.

Contents¶

Exceptions¶

Salt-specific exceptions should be thrown as often as possible so the various interfaces to Salt (CLI, API, etc) can handle those errors appropriately and display error messages appropriately.

salt.exceptions This module is a central location for all salt exceptions

Network Topology¶

Salt is based on a powerful, asynchronous, network topology using ZeroMQ. Many ZeroMQ systems are in place to enable communication. The central idea is to have the fastest communication possible.

Servers¶

The Salt Master runs 2 network services. First is the ZeroMQ PUB system. This service by default runs on port 4505 and can be configured via the publish_port option in the master configuration.

Second is the ZeroMQ REP system. This is a separate interface used for all bi-directional communication with minions. By default this system binds to port 4506 and can be configured via the ret_port option in the master.

PUB/SUB¶

The commands sent out via the salt client are broadcast out to the minions via ZeroMQ PUB/SUB. This is done by allowing the minions to maintain a connection back to the Salt Master and then all connections are informed to download the command data at once. The command data is kept extremely small (usually less than 1K) so it is not a burden on the network.

Return¶

The PUB/SUB system is a one way communication, so once a publish is sent out the PUB interface on the master has no further communication with the minion. The minion, after running the command, then sends the command's return data back to the master via the ret_port.

Windows Software Repository¶

The Salt Windows Software Repository provides a package manager and software repository similar to what is provided by yum and apt on Linux.

It permits the installation of software using the installers on remote windows machines. In many senses, the operation is similar to that of the other package managers salt is aware of:

the pkg.installed and similar states work on Windows.
the pkg.install and similar module functions work on Windows.
each windows machine needs to have pkg.refresh_db executed against it to pick up the latest version of the package database.

High level differences to yum and apt are:

The repository metadata (sls files) is hosted through either salt or git.
Packages can be downloaded from within the salt repository, a git repository or from http(s) or ftp urls.
No dependencies are managed. Dependencies between packages needs to be managed manually.

Operation¶

The install state/module function of the windows package manager works roughly as follows:

Execute pkg.list_pkgs and store the result
Check if any action needs to be taken. (ie compare required package and version against pkg.list_pkgs results)
If so, run the installer command.
Execute pkg.list_pkgs and compare to the result stored from before installation.
Sucess/Failure/Changes will be reported based on the differences between the original and final pkg.list_pkgs results.

If there are any problems in using the package manager it is likely to be due to the data in your sls files not matching the difference between the pre and post pkg.list_pkgs results.

Usage¶

By default, the Windows software repository is found at /srv/salt/win/repo This can be changed in the master config file (default location is /etc/salt/master) by modifying the win_repo variable. Each piece of software should have its own directory which contains the installers and a package definition file. This package definition file is a YAML file named init.sls.

The package definition file should look similar to this example for Firefox: /srv/salt/win/repo/firefox/init.sls

firefox:
  17.0.1:
    installer: 'salt://win/repo/firefox/English/Firefox Setup 17.0.1.exe'
    full_name: Mozilla Firefox 17.0.1 (x86 en-US)
    locale: en_US
    reboot: False
    install_flags: ' -ms'
    uninstaller: '%ProgramFiles(x86)%/Mozilla Firefox/uninstall/helper.exe'
    uninstall_flags: ' /S'
  16.0.2:
    installer: 'salt://win/repo/firefox/English/Firefox Setup 16.0.2.exe'
    full_name: Mozilla Firefox 16.0.2 (x86 en-US)
    locale: en_US
    reboot: False
    install_flags: ' -ms'
    uninstaller: '%ProgramFiles(x86)%/Mozilla Firefox/uninstall/helper.exe'
    uninstall_flags: ' /S'
  15.0.1:
    installer: 'salt://win/repo/firefox/English/Firefox Setup 15.0.1.exe'
    full_name: Mozilla Firefox 15.0.1 (x86 en-US)
    locale: en_US
    reboot: False
    install_flags: ' -ms'
    uninstaller: '%ProgramFiles(x86)%/Mozilla Firefox/uninstall/helper.exe'
    uninstall_flags: ' /S'

More examples can be found here: https://github.com/saltstack/salt-winrepo

The version number and full_name need to match the output from pkg.list_pkgs so that the status can be verfied when running highstate. Note: It is still possible to successfully install packages using pkg.install even if they don't match which can make this hard to troubleshoot.

salt 'test-2008' pkg.list_pkgs
test-2008
    ----------
    7-Zip 9.20 (x64 edition):
        9.20.00.0
    Microsoft .NET Framework 4 Client Profile:
        4.0.30319,4.0.30319
    Microsoft .NET Framework 4 Extended:
        4.0.30319,4.0.30319
    Microsoft Visual C++ 2008 Redistributable - x64 9.0.21022:
        9.0.21022
    Mozilla Firefox 17.0.1 (x86 en-US):
        17.0.1
    Mozilla Maintenance Service:
        17.0.1
    NSClient++ (x64):
        0.3.8.76
    Notepad++:
        6.4.2
    Salt Minion 0.16.0:
        0.16.0

If any of these preinstalled packages already exist in winrepo the full_name will be automatically renamed to their package name during the next update (running highstate or installing another package).

test-2008:
    ----------
    7zip:
        9.20.00.0
    Microsoft .NET Framework 4 Client Profile:
        4.0.30319,4.0.30319
    Microsoft .NET Framework 4 Extended:
        4.0.30319,4.0.30319
    Microsoft Visual C++ 2008 Redistributable - x64 9.0.21022:
        9.0.21022
    Mozilla Maintenance Service:
        17.0.1
    Notepad++:
        6.4.2
    Salt Minion 0.16.0:
        0.16.0
    firefox:
        17.0.1
    nsclient:
        0.3.9.328

Add msiexec: True if using an MSI installer requiring the use of msiexec /i to install and msiexec /x to uninstall.

The install_flags and uninstall_flags are flags passed to the software installer to cause it to perform a silent install. These can often be found by adding /? or /h when running the installer from the command line. A great resource for finding these silent install flags can be found on the WPKG project's wiki:

7zip:
  9.20.00.0:
    installer: salt://win/repo/7zip/7z920-x64.msi
    full_name: 7-Zip 9.20 (x64 edition)
    reboot: False
    install_flags: ' /q '
    msiexec: True
    uninstaller: salt://win/repo/7zip/7z920-x64.msi
    uninstall_flags: ' /qn'

Generate Repo Cache File¶

Once the sls file has been created, generate the repository cache file with the winrepo runner:

salt-run winrepo.genrepo

Then update the repository cache file on your minions, exactly how it's done for the Linux package managers:

salt '*' pkg.refresh_db

Install Windows Software¶

Now you can query the available version of Firefox using the Salt pkg module.

salt '*' pkg.available_version firefox

{'davewindows': {'15.0.1': 'Mozilla Firefox 15.0.1 (x86 en-US)',
                 '16.0.2': 'Mozilla Firefox 16.0.2 (x86 en-US)',
                 '17.0.1': 'Mozilla Firefox 17.0.1 (x86 en-US)'}}

As you can see, there are three versions of Firefox available for installation.

salt '*' pkg.install firefox

The above line will install the latest version of Firefox.

salt '*' pkg.install firefox version=16.0.2

The above line will install version 16.0.2 of Firefox.

If a different version of the package is already installed it will be replaced with the version in winrepo (only if the package itself supports live updating)

Uninstall Windows Software¶

Uninstall software using the pkg module:

salt '*' pkg.remove firefox

salt '*' pkg.purge firefox

pkg.purge just executes pkg.remove on Windows. At some point in the future pkg.purge may direct the installer to remove all configs and settings for software packages that support that option.

Standalone Minion Salt Windows Repo Module¶

In order to facilitate managing a Salt Windows software repo with Salt on a Standalone Minion on Windows, a new module named winrepo has been added to Salt. wirepo matches what is available in the salt runner and allows you to manage the Windows software repo contents. Example: salt '*' winrepo.genrepo

Git Hosted Repo¶

Windows software package definitions can also be hosted in one or more git repositories. The default repo is one hosted on Github.com by SaltStack,Inc., which includes package definitions for open source software. This repo points to the HTTP or ftp locations of the installer files. Anyone is welcome to send a pull request to this repo to add new package definitions. Browse the repo here: https://github.com/saltstack/salt-winrepo .

Configure which git repos the master can search for package definitions by modifying or extending the win_gitrepos configuration option list in the master config.

Checkout each git repo in win_gitrepos, compile your package repository cache and then refresh each minion's package cache:

salt-run winrepo.update_git_repos
salt-run winrepo.genrepo
salt '*' pkg.refresh_db

Troubleshooting¶

Incorrect name/version¶

If the package seems to install properly, but salt reports a failure then it is likely you have a version or full_name mismatch.

Check the exact full_name and version used by the package. Use pkg.list_pkgs to check that the names and version exactly match what is installed.

Changes to sls files not being picked up¶

Ensure you have (re)generated the repository cache file and then updated the repository cache on the relevant minions:

salt-run winrepo.genrepo
salt 'MINION' pkg.refresh_db

Packages management under Windows 2003¶

On windows server 2003, you need to install optional windows component "wmi windows installer provider" to have full list of installed packages. If you don't have this, salt-minion can't report some installed software.

Command Line Reference¶

Salt can be controlled by a command line client by the root user on the Salt master. The Salt command line client uses the Salt client API to communicate with the Salt master server. The Salt client is straightforward and simple to use.

Using the Salt client commands can be easily sent to the minions.

Each of these commands accepts an explicit --config option to point to either the master or minion configuration file. If this option is not provided and the default configuration file does not exist then Salt falls back to use the environment variables SALT_MASTER_CONFIG and SALT_MINION_CONFIG.

See also

Configuring Salt

Using the Salt Command¶

The Salt command needs a few components to send information to the Salt minions. The target minions need to be defined, the function to call and any arguments the function requires.

Defining the Target Minions¶

The first argument passed to salt, defines the target minions, the target minions are accessed via their hostname. The default target type is a bash glob:

salt '*foo.com' sys.doc

Salt can also define the target minions with regular expressions:

salt -E '.*' cmd.run 'ls -l | grep foo'

Or to explicitly list hosts, salt can take a list:

salt -L foo.bar.baz,quo.qux cmd.run 'ps aux | grep foo'

More Powerful Targets¶

The simple target specifications, glob, regex and list will cover many use cases, and for some will cover all use cases, but more powerful options exist.

Targeting with Grains¶

The Grains interface was built into Salt to allow minions to be targeted by system properties. So minions running on a particular operating system can be called to execute a function, or a specific kernel.

Calling via a grain is done by passing the -G option to salt, specifying a grain and a glob expression to match the value of the grain. The syntax for the target is the grain key followed by a globexpression: "os:Arch*".

salt -G 'os:Fedora' test.ping

Will return True from all of the minions running Fedora.

To discover what grains are available and what the values are, execute the grains.item salt function:

salt '*' grains.items

Targeting with Executions¶

As of 0.8.8 targeting with executions is still under heavy development and this documentation is written to reference the behavior of execution matching in the future.

Execution matching allows for a primary function to be executed, and then based on the return of the primary function the main function is executed.

Execution matching allows for matching minions based on any arbitrary running data on the minions.

Compound Targeting¶

New in version 0.9.5.

Multiple target interfaces can be used in conjunction to determine the command targets. These targets can then be combined using and or or statements. This is well defined with an example:

salt -C 'G@os:Debian and webser* or E@db.*' test.ping

In this example any minion who's id starts with webser and is running Debian, or any minion who's id starts with db will be matched.

The type of matcher defaults to glob, but can be specified with the corresponding letter followed by the @ symbol. In the above example a grain is used with G@ as well as a regular expression with E@. The webser* target does not need to be prefaced with a target type specifier because it is a glob.

Node Group Targeting¶

New in version 0.9.5.

Often the convenience of having a predefined group of minions to execute targets on is desired. This can be accomplished with the new nodegroups feature. Nodegroups allow for predefined compound targets to be declared in the master configuration file:

nodegroups:
  group1: 'L@foo.domain.com,bar.domain.com,baz.domain.com and bl*.domain.com'
  group2: 'G@os:Debian and foo.domain.com'

Calling the Function¶

The function to call on the specified target is placed after the target specification.

New in version 0.9.8.

Functions may also accept arguments, space-delimited:

salt '*' cmd.exec_code python 'import sys; print sys.version'

Optional, keyword arguments are also supported:

salt '*' pip.install salt timeout=5 upgrade=True

They are always in the form of kwarg=argument.

Arguments are formatted as YAML:

salt '*' cmd.run 'echo "Hello: $FIRST_NAME"' env='{FIRST_NAME: "Joe"}'

Note: dictionaries must have curly braces around them (like the env keyword argument above). This was changed in 0.15.1: in the above example, the first argument used to be parsed as the dictionary {'echo "Hello': '$FIRST_NAME"'}. This was generally not the expected behavior.

If you want to test what parameters are actually passed to a module, use the test.arg_repr command:

salt '*' test.arg_repr 'echo "Hello: $FIRST_NAME"' env='{FIRST_NAME: "Joe"}'

Finding available minion functions¶

The Salt functions are self documenting, all of the function documentation can be retried from the minions via the sys.doc() function:

salt '*' sys.doc

Compound Command Execution¶

If a series of commands needs to be sent to a single target specification then the commands can be sent in a single publish. This can make gathering groups of information faster, and lowers the stress on the network for repeated commands.

Compound command execution works by sending a list of functions and arguments instead of sending a single function and argument. The functions are executed on the minion in the order they are defined on the command line, and then the data from all of the commands are returned in a dictionary. This means that the set of commands are called in a predictable way, and the returned data can be easily interpreted.

Executing compound commands if done by passing a comma delimited list of functions, followed by a comma delimited list of arguments:

salt '*' cmd.run,test.ping,test.echo 'cat /proc/cpuinfo',,foo

The trick to look out for here, is that if a function is being passed no arguments, then there needs to be a placeholder for the absent arguments. This is why in the above example, there are two commas right next to each other. test.ping takes no arguments, so we need to add another comma, otherwise Salt would attempt to pass "foo" to test.ping.

If you need to pass arguments that include commas, then make sure you add spaces around the commas that separate arguments. For example:

salt '*' cmd.run,test.ping,test.echo 'echo "1,2,3"' , , foo

You may change the arguments separator using the --args-separator option:

salt --args-separator=:: '*' some.fun,test.echo params with , comma :: foo

`salt`¶

Synopsis¶

salt '*' [ options ] sys.doc

salt -E '.*' [ options ] sys.doc cmd

salt -G 'os:Arch.*' [ options ] test.ping

salt -C 'G@os:Arch.* and webserv* or G@kernel:FreeBSD' [ options ] test.ping

Description¶

Salt allows for commands to be executed across a swath of remote systems in parallel. This means that remote systems can be both controlled and queried with ease.

Options¶

--version¶: Print the version of Salt that is running.

--versions-report¶: Show program's dependencies and version number, and then exit

-h, --help¶: Show the help message and exit

-c CONFIG_DIR, --config-dir=CONFIG_dir¶: The location of the Salt configuration directory. This directory contains the configuration files for Salt master and minions. The default location on most systems is /etc/salt.

-t TIMEOUT, --timeout=TIMEOUT¶: The timeout in seconds to wait for replies from the Salt minions. The timeout number specifies how long the command line client will wait to query the minions and check on running jobs. Default: 5

-s, --static¶: By default as of version 0.9.8 the salt command returns data to the console as it is received from minions, but previous releases would return data only after all data was received. To only return the data with a hard timeout and after all minions have returned then use the static option.

--async¶: Instead of waiting for the job to run on minions only print the jod id of the started execution and complete.

--state-output=STATE_OUTPUT¶: New in version 0.17.

Override the configured state_output value for minion output. Default: full

--subset=SUBSET¶: Execute the routine on a random subset of the targeted minions. The minions will be verified that they have the named function before executing.

-v VERBOSE, --verbose¶: Turn on verbosity for the salt call, this will cause the salt command to print out extra data like the job id.

-b BATCH, --batch-size=BATCH¶: Instead of executing on all targeted minions at once, execute on a progressive set of minions. This option takes an argument in the form of an explicit number of minions to execute at once, or a percentage of minions to execute on.

-a EAUTH, --auth=EAUTH¶: Pass in an external authentication medium to validate against. The credentials will be prompted for. Can be used with the -T option.

-T, --make-token¶: Used in conjunction with the -a option. This creates a token that allows for the authenticated user to send commands without needing to re-authenticate.

--return=RETURNER¶: Chose an alternative returner to call on the minion, if an alternative returner is used then the return will not come back to the command line but will be sent to the specified return system.

-d, --doc, --documentation¶: Return the documentation for the module functions available on the minions

--args-separator=ARGS_SEPARATOR¶: Set the special argument used as a delimiter between command arguments of compound commands. This is useful when one wants to pass commas as arguments to some of the commands in a compound command.

Logging Options¶

Logging options which override any settings defined on the configuration files.

-l LOG_LEVEL, --log-level=LOG_LEVEL¶: Console logging log level. One of all, garbage, trace, debug, info, warning, error, quiet. Default: warning.

--log-file=LOG_FILE¶: Log file path. Default: /var/log/salt/master.

--log-file-level=LOG_LEVEL_LOGFILE¶: Logfile logging log level. One of all, garbage, trace, debug, info, warning, error, quiet. Default: warning.

Target Selection¶

-E, --pcre¶: The target expression will be interpreted as a PCRE regular expression rather than a shell glob.

-L, --list¶: The target expression will be interpreted as a comma-delimited list; example: server1.foo.bar,server2.foo.bar,example7.quo.qux

-G, --grain¶

The target expression matches values returned by the Salt grains system on the minions. The target expression is in the format of '<grain value>:<glob expression>'; example: 'os:Arch*'

This was changed in version 0.9.8 to accept glob expressions instead of regular expression. To use regular expression matching with grains, use the --grain-pcre option.

--grain-pcre¶: The target expression matches values returned by the Salt grains system on the minions. The target expression is in the format of '<grain value>:< regular expression>'; example: 'os:Arch.*'

-N, --nodegroup¶: Use a predefined compound target defined in the Salt master configuration file.

-R, --range¶

Instead of using shell globs to evaluate the target, use a range expression to identify targets. Range expressions look like %cluster.

Using the Range option requires that a range server is set up and the location of the range server is referenced in the master configuration file.

-C, --compound¶: Utilize many target definitions to make the call very granular. This option takes a group of targets separated by and or or. The default matcher is a glob as usual. If something other than a glob is used, preface it with the letter denoting the type; example: 'webserv* and G@os:Debian or E@db*' Make sure that the compound target is encapsulated in quotes.

-X, --exsel¶: Instead of using shell globs, use the return code of a function.

-I, --pillar¶: Instead of using shell globs to evaluate the target, use a pillar value to identify targets. The syntax for the target is the pillar key followed by a glob expression: "role:production*"

-S, --ipcidr¶: Match based on Subnet (CIDR notation) or IPv4 address.

Output Options¶

--out¶

Pass in an alternative outputter to display the return of data. This outputter can be any of the available outputters:

grains, highstate, json, key, overstatestage, pprint, raw, txt, yaml

Some outputters are formatted only for data returned from specific functions; for instance, the grains outputter will not work for non-grains data.

If an outputter is used that does not support the data passed into it, then Salt will fall back on the pprint outputter and display the return data using the Python pprint standard library module.

Note

If using --out=json, you will probably want --static as well. Without the static option, you will get a JSON string for each minion. This is due to using an iterative outputter. So if you want to feed it to a JSON parser, use --static as well.

--out-indent OUTPUT_INDENT, --output-indent OUTPUT_INDENT¶: Print the output indented by the provided value in spaces. Negative values disable indentation. Only applicable in outputters that support indentation.

--out-file=OUTPUT_FILE, --output-file=OUTPUT_FILE¶: Write the output to the specified file.

--no-color¶: Disable all colored output

--force-color¶: Force colored output

`salt-master`¶

The Salt master daemon, used to control the Salt minions

Synopsis¶

salt-master [ options ]

Description¶

The master daemon controls the Salt minions

Options¶

--version¶: Print the version of Salt that is running.

--versions-report¶: Show program's dependencies and version number, and then exit

-h, --help¶: Show the help message and exit

-c CONFIG_DIR, --config-dir=CONFIG_dir¶: The location of the Salt configuration directory. This directory contains the configuration files for Salt master and minions. The default location on most systems is /etc/salt.

-u USER, --user=USER¶: Specify user to run salt-master

-d, --daemon¶: Run salt-master as a daemon

--pid-file PIDFILE¶: Specify the location of the pidfile. Default: /var/run/salt-master.pid

Logging Options¶

Logging options which override any settings defined on the configuration files.

-l LOG_LEVEL, --log-level=LOG_LEVEL¶: Console logging log level. One of all, garbage, trace, debug, info, warning, error, quiet. Default: warning.

--log-file=LOG_FILE¶: Log file path. Default: /var/log/salt/master.

--log-file-level=LOG_LEVEL_LOGFILE¶: Logfile logging log level. One of all, garbage, trace, debug, info, warning, error, quiet. Default: warning.

`salt-minion`¶

The Salt minion daemon, receives commands from a remote Salt master.

Synopsis¶

salt-minion [ options ]

Description¶

The Salt minion receives commands from the central Salt master and replies with the results of said commands.

Options¶

--version¶: Print the version of Salt that is running.

--versions-report¶: Show program's dependencies and version number, and then exit

-h, --help¶: Show the help message and exit

-c CONFIG_DIR, --config-dir=CONFIG_dir¶: The location of the Salt configuration directory. This directory contains the configuration files for Salt master and minions. The default location on most systems is /etc/salt.

-u USER, --user=USER¶: Specify user to run salt-minion

-d, --daemon¶: Run salt-minion as a daemon

--pid-file PIDFILE¶: Specify the location of the pidfile. Default: /var/run/salt-minion.pid

Logging Options¶

Logging options which override any settings defined on the configuration files.

-l LOG_LEVEL, --log-level=LOG_LEVEL¶: Console logging log level. One of all, garbage, trace, debug, info, warning, error, quiet. Default: warning.

--log-file=LOG_FILE¶: Log file path. Default: /var/log/salt/minion.

--log-file-level=LOG_LEVEL_LOGFILE¶: Logfile logging log level. One of all, garbage, trace, debug, info, warning, error, quiet. Default: warning.

`salt-key`¶

Synopsis¶

salt-key [ options ]

Description¶

Salt-key executes simple management of Salt server public keys used for authentication.

Options¶

--version¶: Print the version of Salt that is running.

--versions-report¶: Show program's dependencies and version number, and then exit

-h, --help¶: Show the help message and exit

-c CONFIG_DIR, --config-dir=CONFIG_dir¶: The location of the Salt configuration directory. This directory contains the configuration files for Salt master and minions. The default location on most systems is /etc/salt.

-q, --quiet¶: Suppress output

-y, --yes¶: Answer 'Yes' to all questions presented, defaults to False

Logging Options¶

Logging options which override any settings defined on the configuration files.

--log-file=LOG_FILE¶: Log file path. Default: /var/log/salt/minion.

--log-file-level=LOG_LEVEL_LOGFILE¶: Logfile logging log level. One of all, garbage, trace, debug, info, warning, error, quiet. Default: warning.

Output Options¶

--out¶

Pass in an alternative outputter to display the return of data. This outputter can be any of the available outputters:

grains, highstate, json, key, overstatestage, pprint, raw, txt, yaml

Some outputters are formatted only for data returned from specific functions; for instance, the grains outputter will not work for non-grains data.

If an outputter is used that does not support the data passed into it, then Salt will fall back on the pprint outputter and display the return data using the Python pprint standard library module.

Note

If using --out=json, you will probably want --static as well. Without the static option, you will get a JSON string for each minion. This is due to using an iterative outputter. So if you want to feed it to a JSON parser, use --static as well.

--out-indent OUTPUT_INDENT, --output-indent OUTPUT_INDENT¶: Print the output indented by the provided value in spaces. Negative values disable indentation. Only applicable in outputters that support indentation.

--out-file=OUTPUT_FILE, --output-file=OUTPUT_FILE¶: Write the output to the specified file.

--no-color¶: Disable all colored output

--force-color¶: Force colored output

Actions¶

-l ARG, --list=ARG¶: List the public keys. The args "pre", "un", and "unaccepted" will list unaccepted/unsigned keys. "acc" or "accepted" will list accepted/signed keys. "rej" or "rejected" will list rejected keys. Finally, "all" will list all keys.

-L, --list-all¶: List all public keys on this Salt master: accepted, pending, and rejected.

-a ACCEPT, --accept=ACCEPT¶: Accept the named minion public key for command execution.

-A, --accept-all¶: Accepts all pending public keys.

-r REJECT, --reject=REJECT¶: Reject the named minion public key.

-R, --reject-all¶: Rejects all pending public keys.

-p PRINT, --print=PRINT¶: Print the specified public key

-P, --print-all¶: Print all public keys

-d DELETE, --delete=DELETE¶: Delete the named minion key or minion keys matching a glob for command execution.

-D, --delete-all¶: Delete all keys

-f FINGER, --finger=FINGER¶: Print the named key's fingerprint

-F, --finger-all¶: Print all key's fingerprints

Key Generation Options¶

--gen-keys=GEN_KEYS¶: Set a name to generate a keypair for use with salt

--gen-keys-dir=GEN_KEYS_DIR¶: Set the directory to save the generated keypair. Only works with 'gen_keys_dir' option; default is the current directory.

--keysize=KEYSIZE¶: Set the keysize for the generated key, only works with the '--gen-keys' option, the key size must be 2048 or higher, otherwise it will be rounded up to 2048. The default is 2048.

`salt-cp`¶

Copy a file to a set of systems

Synopsis¶

salt-cp '*' [ options ] SOURCE DEST

salt-cp -E '.*' [ options ] SOURCE DEST

salt-cp -G 'os:Arch.*' [ options ] SOURCE DEST

Description¶

Salt copy copies a local file out to all of the Salt minions matched by the given target.

Options¶

--version¶: Print the version of Salt that is running.

--versions-report¶: Show program's dependencies and version number, and then exit

-h, --help¶: Show the help message and exit

-c CONFIG_DIR, --config-dir=CONFIG_dir¶: The location of the Salt configuration directory. This directory contains the configuration files for Salt master and minions. The default location on most systems is /etc/salt.

-t TIMEOUT, --timeout=TIMEOUT¶: The timeout in seconds to wait for replies from the Salt minions. The timeout number specifies how long the command line client will wait to query the minions and check on running jobs. Default: 5

Logging Options¶

Logging options which override any settings defined on the configuration files.

-l LOG_LEVEL, --log-level=LOG_LEVEL¶: Console logging log level. One of all, garbage, trace, debug, info, warning, error, quiet. Default: warning.

--log-file=LOG_FILE¶: Log file path. Default: /var/log/salt/master.

--log-file-level=LOG_LEVEL_LOGFILE¶: Logfile logging log level. One of all, garbage, trace, debug, info, warning, error, quiet. Default: warning.

Target Selection¶

-E, --pcre¶: The target expression will be interpreted as a PCRE regular expression rather than a shell glob.

-L, --list¶: The target expression will be interpreted as a comma-delimited list; example: server1.foo.bar,server2.foo.bar,example7.quo.qux

-G, --grain¶

The target expression matches values returned by the Salt grains system on the minions. The target expression is in the format of '<grain value>:<glob expression>'; example: 'os:Arch*'

This was changed in version 0.9.8 to accept glob expressions instead of regular expression. To use regular expression matching with grains, use the --grain-pcre option.

--grain-pcre¶: The target expression matches values returned by the Salt grains system on the minions. The target expression is in the format of '<grain value>:< regular expression>'; example: 'os:Arch.*'

-N, --nodegroup¶: Use a predefined compound target defined in the Salt master configuration file.

-R, --range¶

Instead of using shell globs to evaluate the target, use a range expression to identify targets. Range expressions look like %cluster.

Using the Range option requires that a range server is set up and the location of the range server is referenced in the master configuration file.

`salt-call`¶

Synopsis¶

salt-call [options]

Description¶

The salt-call command is used to run module functions locally on a minion instead of executing them from the master.

Options¶

--version¶: Print the version of Salt that is running.

--versions-report¶: Show program's dependencies and version number, and then exit

-h, --help¶: Show the help message and exit

-c CONFIG_DIR, --config-dir=CONFIG_dir¶: The location of the Salt configuration directory. This directory contains the configuration files for Salt master and minions. The default location on most systems is /etc/salt.

-g, --grains¶: Return the information generated by the Salt grains

-m MODULE_DIRS, --module-dirs=MODULE_DIRS¶: Specify an additional directories to pull modules from, multiple directories can be delimited by commas

-d, --doc, --documentation¶: Return the documentation for the specified module or for all modules if none are specified

--master=MASTER¶: Specify the master to use. The minion must be authenticated with the master. If this option is omitted, the master options from the minion config will be used. If multi masters are set up the first listed master that responds will be used.

--return RETURNER¶: Set salt-call to pass the return data to one or many returner interfaces. To use many returner interfaces specify a comma delimited list of returners.

--local¶: Run salt-call locally, as if there was no master running.

Logging Options¶

Logging options which override any settings defined on the configuration files.

-l LOG_LEVEL, --log-level=LOG_LEVEL¶: Console logging log level. One of all, garbage, trace, debug, info, warning, error, quiet. Default: info.

--log-file=LOG_FILE¶: Log file path. Default: /var/log/salt/minion.

--log-file-level=LOG_LEVEL_LOGFILE¶: Logfile logging log level. One of all, garbage, trace, debug, info, warning, error, quiet. Default: info.

Output Options¶

--out¶

Pass in an alternative outputter to display the return of data. This outputter can be any of the available outputters:

grains, highstate, json, key, overstatestage, pprint, raw, txt, yaml

Some outputters are formatted only for data returned from specific functions; for instance, the grains outputter will not work for non-grains data.

If an outputter is used that does not support the data passed into it, then Salt will fall back on the pprint outputter and display the return data using the Python pprint standard library module.

Note

If using --out=json, you will probably want --static as well. Without the static option, you will get a JSON string for each minion. This is due to using an iterative outputter. So if you want to feed it to a JSON parser, use --static as well.

--out-indent OUTPUT_INDENT, --output-indent OUTPUT_INDENT¶: Print the output indented by the provided value in spaces. Negative values disable indentation. Only applicable in outputters that support indentation.

--out-file=OUTPUT_FILE, --output-file=OUTPUT_FILE¶: Write the output to the specified file.

--no-color¶: Disable all colored output

--force-color¶: Force colored output

`salt-run`¶

Execute a Salt runner

Synopsis¶

salt-run RUNNER

Description¶

salt-run is the frontend command for executing Salt Runners. Salt runners are simple modules used to execute convenience functions on the master

Options¶

--version¶: Print the version of Salt that is running.

--versions-report¶: Show program's dependencies and version number, and then exit

-h, --help¶: Show the help message and exit

-c CONFIG_DIR, --config-dir=CONFIG_dir¶: The location of the Salt configuration directory. This directory contains the configuration files for Salt master and minions. The default location on most systems is /etc/salt.

-t TIMEOUT, --timeout=TIMEOUT¶: The timeout in seconds to wait for replies from the Salt minions. The timeout number specifies how long the command line client will wait to query the minions and check on running jobs. Default: 1

-d, --doc, --documentation¶: Display documentation for runners, pass a module or a runner to see documentation on only that module/runner.

Logging Options¶

Logging options which override any settings defined on the configuration files.

-l LOG_LEVEL, --log-level=LOG_LEVEL¶: Console logging log level. One of all, garbage, trace, debug, info, warning, error, quiet. Default: warning.

--log-file=LOG_FILE¶: Log file path. Default: /var/log/salt/master.

--log-file-level=LOG_LEVEL_LOGFILE¶: Logfile logging log level. One of all, garbage, trace, debug, info, warning, error, quiet. Default: warning.

`salt-ssh`¶

Synopsis¶

salt-ssh '*' [ options ] sys.doc

salt-ssh -E '.*' [ options ] sys.doc cmd

Description¶

Salt ssh allows for salt routines to be executed using only ssh for transport

Options¶

--version¶: Print the version of Salt that is running.

--versions-report¶: Show program's dependencies and version number, and then exit

-h, --help¶: Show the help message and exit

-c CONFIG_DIR, --config-dir=CONFIG_dir¶: The location of the Salt configuration directory. This directory contains the configuration files for Salt master and minions. The default location on most systems is /etc/salt.

Target Selection¶

-E, --pcre¶: The target expression will be interpreted as a PCRE regular expression rather than a shell glob.

-L, --list¶: The target expression will be interpreted as a comma-delimited list; example: server1.foo.bar,server2.foo.bar,example7.quo.qux

-G, --grain¶

The target expression matches values returned by the Salt grains system on the minions. The target expression is in the format of '<grain value>:<glob expression>'; example: 'os:Arch*'

This was changed in version 0.9.8 to accept glob expressions instead of regular expression. To use regular expression matching with grains, use the --grain-pcre option.

--grain-pcre¶: The target expression matches values returned by the Salt grains system on the minions. The target expression is in the format of '<grain value>:< regular expression>'; example: 'os:Arch.*'

-N, --nodegroup¶: Use a predefined compound target defined in the Salt master configuration file.

-R, --range¶

Instead of using shell globs to evaluate the target, use a range expression to identify targets. Range expressions look like %cluster.

Using the Range option requires that a range server is set up and the location of the range server is referenced in the master configuration file.

Logging Options¶

Logging options which override any settings defined on the configuration files.

-l LOG_LEVEL, --log-level=LOG_LEVEL¶: Console logging log level. One of all, garbage, trace, debug, info, warning, error, quiet. Default: warning.

--log-file=LOG_FILE¶: Log file path. Default: /var/log/salt/ssh.

--log-file-level=LOG_LEVEL_LOGFILE¶: Logfile logging log level. One of all, garbage, trace, debug, info, warning, error, quiet. Default: warning.

Output Options¶

--out¶

Pass in an alternative outputter to display the return of data. This outputter can be any of the available outputters:

grains, highstate, json, key, overstatestage, pprint, raw, txt, yaml

Some outputters are formatted only for data returned from specific functions; for instance, the grains outputter will not work for non-grains data.

If an outputter is used that does not support the data passed into it, then Salt will fall back on the pprint outputter and display the return data using the Python pprint standard library module.

Note

If using --out=json, you will probably want --static as well. Without the static option, you will get a JSON string for each minion. This is due to using an iterative outputter. So if you want to feed it to a JSON parser, use --static as well.

--out-indent OUTPUT_INDENT, --output-indent OUTPUT_INDENT¶: Print the output indented by the provided value in spaces. Negative values disable indentation. Only applicable in outputters that support indentation.

--out-file=OUTPUT_FILE, --output-file=OUTPUT_FILE¶: Write the output to the specified file.

--no-color¶: Disable all colored output

--force-color¶: Force colored output

`salt-syndic`¶

The Salt syndic daemon, a special minion that passes through commands from a higher master

Synopsis¶

salt-syndic [ options ]

Description¶

The Salt syndic daemon, a special minion that passes through commands from a higher master.

Options¶

--version¶: Print the version of Salt that is running.

--versions-report¶: Show program's dependencies and version number, and then exit

-h, --help¶: Show the help message and exit

-c CONFIG_DIR, --config-dir=CONFIG_dir¶: The location of the Salt configuration directory. This directory contains the configuration files for Salt master and minions. The default location on most systems is /etc/salt.

-u USER, --user=USER¶: Specify user to run salt-syndic

-d, --daemon¶: Run salt-syndic as a daemon

--pid-file PIDFILE¶: Specify the location of the pidfile. Default: /var/run/salt-syndic.pid

Logging Options¶

Logging options which override any settings defined on the configuration files.

-l LOG_LEVEL, --log-level=LOG_LEVEL¶: Console logging log level. One of all, garbage, trace, debug, info, warning, error, quiet. Default: warning.

--log-file=LOG_FILE¶: Log file path. Default: /var/log/salt/master.

--log-file-level=LOG_LEVEL_LOGFILE¶: Logfile logging log level. One of all, garbage, trace, debug, info, warning, error, quiet. Default: warning.

Release notes and upgrade instructions¶

Salt 0.10.0 Release Notes¶

0.10.0 has arrived! This release comes with MANY bug fixes, and new capabilities which greatly enhance performance and reliability. This release is primarily a bug fix release with many new tests and many repaired bugs. This release also introduces a few new key features which were brought in primarily to repair bugs and some limitations found in some of the components of the original architecture.

Major Features¶

Event System¶

The Salt Master now comes equipped with a new event system. This event system has replaced some of the back end of the Salt client and offers the beginning of a system which will make plugging external applications into Salt. The event system relies on a local ZeroMQ publish socket and other processes can connect to this socket and listen for events. The new events can be easily managed via Salt's event library.

Unprivileged User Updates¶

Some enhancements have been added to Salt for running as a user other than root. These new additions should make switching the user that the Salt Master is running as very painless, simply change the user option in the master configuration and restart the master, Salt will take care of all of the particulars for you.

Peer Runner Execution¶

Salt has long had the peer communication system used to allow minions to send commands via the salt master. 0.10.0 adds a new capability here, now the master can be configured to allow for minions to execute Salt runners via the peer_run option in the salt master configuration.

YAML Parsing Updates¶

In the past the YAML parser for sls files would return the incorrect numbers when the file mode was set with a preceding 0. The YAML parser used in Salt has been modified to no longer convert these number into octal but to keep them as the correct value so that sls files can be a little cleaner to write.

State Call Data Files¶

It was requested that the minion keep a local cache of the most recent executed state run. This has been added and now with state runs the data is stored in a msgpack file in the minion's cachedir.

Turning Off the Job Cache¶

A new option has been added to the master configuration file. In previous releases the Salt client would look over the Salt job cache to read in the minion return data. With the addition of the event system the Salt client can now watch for events directly from the master worker processes.

This means that the job cache is no longer a hard requirement. Keep in mind though, that turning off the job cache means that historic job execution data cannot be retrieved.

Test Updates¶

Minion Swarms Are Faster¶

To continue our efforts with testing Salt's ability to scale the minionswarm script has been updated. The minionswarm can now start up minions much faster than it could before and comes with a new feature allowing modules to be disabled, thus lowering the minion's footprint when making a swarm. These new updates have allows us to test

# python minionswarm.py -m 20 --master salt-master

Many Fixes¶

To get a good idea for the number of bugfixes this release offers take a look at the closed tickets for 0.10.0, this is a very substantial update:

https://github.com/saltstack/salt/issues?milestone=12&state=closed

Master and Minion Stability Fixes¶

As Salt deployments grow new ways to break Salt are discovered. 0.10.0 comes with a number of fixes for the minions and master greatly improving Salt stability.

Salt 0.10.2 Release Notes¶

0.10.2 is out! This release comes with enhancements to the pillar interface, cleaner ways to access the salt-call capabilities in the API, minion data caching and the event system has been added to salt minions.

There have also been updates to the ZeroMQ functions, many more tests (thanks to sponsors, the code sprint and many contributors) and a swath of bug fixes.

Major Features¶

Ext Pillar Modules¶

The ranks of available Salt modules directories sees a new member in 0.10.2. With the popularity of pillar a higher demand has arisen for ext_pillar interfaces to be more like regular Salt module additions. Now ext_pillar interfaces can be added in the same way as other modules, just drop it into the pillar directory in the salt source.

Minion Events¶

In 0.10.0 an event system was added to the Salt master. 0.10.2 adds the event system to the minions as well. Now event can be published on a local minion as well.

The minions can also send events back up to the master. This means that Salt is able to communicate individual events from the minions back up to the Master which are not associated with command.

Minion Data Caching¶

When pillar was introduced the landscape for available data was greatly enhanced. The minion's began sending grain data back to the master on a regular basis.

The new config option on the master called minion_data_cache instructs the Salt master to maintain a cache of the minion's grains and pillar data in the cachedir. This option is turned off by default to avoid hitting the disk more, but when enabled the cache is used to make grain matching from the salt command more powerful, since the minions that will match can be predetermined.

Backup Files¶

By default all files replaced by the file.managed and file.recurse states we simply deleted. 0.10.2 adds a new option. By setting the backup option to minion the files are backed up before they are replaced.

The backed up files are located in the cachedir under the file_backup directory. On a default system this will be at: /var/cache/salt/file_backup

Configuration files¶

salt-master and salt-minion automatically load additional configuration files from master.d/*.conf respective minion.d/*.conf where master.d/minion.d is a directory in the same directory as the main configuration file.

Salt Key Verification¶

A number of users complained that they had inadvertently deleted the wrong salt authentication keys. 0.10.2 now displays what keys are going to be deleted and verifies that they are the keys that are intended for deletion.

Key auto-signing¶

If autosign_file is specified in the configuration file incoming keys will be compared to the list of keynames in autosign_file. Regular expressions as well as globbing is supported.

The file must only be writable by the user otherwise the file will be ignored. To relax the permission and allow group write access set the permissive_pki_access option.

Module changes¶

Improved OpenBSD support¶

New modules for managing services and packages were provided by Joshua Elsasser to further improve the support for OpenBSD.

Existing modules like the disk module were also improved to support OpenBSD.

SQL Modules¶

The MySQL and PostgreSQL modules have both received a number of additions thanks to the work of Avi Marcus and Roman Imankulov.

ZFS Support on FreeBSD¶

A new ZFS module has been added by Kurtis Velarde for FreeBSD supporting various ZFS operations like creating, extending or removing zpools.

Augeas¶

A new Augeas module by Ulrich Dangel for editing and verifying config files.

Native Debian Service module¶

The support for the Debian was further improved with an new service module for Debian by Ahmad Khayyat supporting disable and enable.

Cassandra¶

Cassandra support has been added by Adam Garside. Currently only status and diagnostic information are supported.

Networking¶

The networking support for RHEL has been improved and supports bonding support as well as zeroconf configuration.

Monit¶

Basic monit support by Kurtis Velarde to control services via monit.

nzbget¶

Basic support for controlling nzbget by Joseph Hall

Bluetooth¶

Baisc bluez support for managing and controlling Bluetooth devices. Supports scanning as well as pairing/unpairing by Joseph Hall.

Test Updates¶

Consistency Testing¶

Another testing script has been added. A bug was found in pillar when many minions generated pillar data at the same time. The new consist.py script is the tests directory was created to reproduce bugs where data should always be consistent.

Many Fixes¶

To get a good idea for the number of bugfixes this release offers take a look at the closed tickets for 0.10.2, this is a very substantial update:

https://github.com/saltstack/salt/issues?milestone=24&page=1&state=closed

Master and Minion Stability Fixes¶

As Salt deployments grow new ways to break Salt are discovered. 0.10.2 comes with a number of fixes for the minions and master greatly improving Salt stability.

Salt 0.10.3 Release Notes¶

The latest taste of Salt has come, this release has many fixes and feature additions. Modifications have been made to make ZeroMQ connections more reliable, the beginning of the ACL system is in place, a new command line parsing system has been added, dynamic module distribution has become more environment aware, the new master_finger option and many more!

Major Features¶

ACL System¶

The new ACL system has been introduced. The ACL system allows for system users other than root to execute salt commands. Users can be allowed to execute specific commands in the same way that minions are opened up to the peer system.

The configuration value to open up the ACL system is called client_acl and is configured like so:

client_acl:
  fred:
    - test..*
    - pkg.list_pkgs

Where fred is allowed access to functions in the test module and to the pkg.list_pkgs function.

Master Finger Option¶

The master_finger option has been added to improve the security of minion provisioning. The master_finger option allows for the fingerprint of the master public key to be set in the configuration file to double verify that the master is valid. This option was added in response to a motivation to pre-authenticate the master when provisioning new minions to help prevent man in the middle attacks in some situations.

Salt Key Fingerprint Generation¶

The ability to generate fingerprints of keys used by Salt has been added to salt-key. The new option finger accepts the name of the key to generate and display a fingerprint for.

salt-key -F master

Will display the fingerprints for the master public and private keys.

Parsing System¶

Pedro Algavio, aka s0undt3ch, has added a substantial update to the command line parsing system that makes the help message output much cleaner and easier to search through. Salt parsers now have --versions-report besides usual --version info which you can provide when reporting any issues found.

Key Generation¶

We have reduced the requirements needed for salt-key to generate minion keys. You're no longer required to have salt configured and it's common directories created just to generate keys. This might prove useful if you're batch creating keys to pre-load on minions.

Startup States¶

A few configuration options have been added which allow for states to be run when the minion daemon starts. This can be a great advantage when deploying with Salt because the minion can apply states right when it first runs. To use startup states set the startup_states configuration option on the minion to highstate.

New Exclude Declaration¶

Some users have asked about adding the ability to ensure that other sls files or ids are excluded from a state run. The exclude statement will delete all of the data loaded from the specified sls file or will delete the specified id:

exclude:
  - sls: http
  - id: /etc/vimrc

Max Open Files¶

While we're currently unable to properly handle ZeroMQ's abort signals when the max open files is reached, due to the way that's handled on ZeroMQ's, we have minimized the chances of this happening without at least warning the user.

More State Output Options¶

Some major changes have been made to the state output system. In the past state return data was printed in a very verbose fashion and only states that failed or made changes were printed by default. Now two options can be passed to the master and minion configuration files to change the behavior of the state output. State output can be set to verbose (default) or non-verbose with the state_verbose option:

state_verbose: False

It is noteworthy that the state_verbose option used to be set to False by default but has been changed to True by default in 0.10.3 due to many requests for the change.

Te next option to be aware of new and called state_output. This option allows for the state output to be set to full (default) or terse.

The full output is the standard state output, but the new terse output will print only one line per state making the output much easier to follow when executing a large state system.

state_output: terse

state.file.append Improvements¶

The salt state file.append() tries not to append existing text. Previously the matching check was being made line by line. While this kind of check might be enough for most cases, if the text being appended was multi-line, the check would not work properly. This issue is now properly handled, the match is done as a whole ignoring any white space addition or removal except inside commas. For those thinking that, in order to properly match over multiple lines, salt will load the whole file into memory, that's not true. For most cases this is not important but an erroneous order to read a 4GB file, if not properly handled, like salt does, could make salt chew that amount of memory. Salt has a buffered file reader which will keep in memory a maximum of 256KB and iterates over the file in chunks of 32KB to test for the match, more than enough, if not, explain your usage on a ticket. With this change, also salt.modules.file.contains(), salt.modules.file.contains_regex(), salt.modules.file.contains_glob() and salt.utils.find now do the searching and/or matching using the buffered chunks approach explained above.

Two new keyword arguments were also added, makedirs and source. The first, makedirs will create the necessary directories in order to append to the specified file, of course, it only applies if we're trying to append to a non-existing file on a non-existing directory:

/tmp/salttest/file-append-makedirs:
    file.append:
        text: foo
        makedirs: True

The second, source, allows one to append the contents of a file instead of specifying the text.

/tmp/salttest/file-append-source:

file.append:
    - source: salt://testfile

Security Fix¶

A timing vulnerability was uncovered in the code which decrypts the AES messages sent over the network. This has been fixed and upgrading is strongly recommended.

Salt 0.10.4 Release Notes¶

Salt 0.10.4 is a monumental release for the Salt team, with two new module systems, many additions to allow granular access to Salt, improved platform support and much more.

This release is also exciting because we have been able to shorten the release cycle back to under a month. We are working hard to keep up the aggressive pace and look forward to having releases happen more frequently!

This release also includes a serious security fix and all users are very strongly recommended to upgrade. As usual, upgrade the master first, and then the minion to ensure that the process is smooth.

Major Features¶

External Authentication System¶

The new external authentication system allows for Salt to pass through authentication to any authentication system to determine if a user has permission to execute a Salt command. The Unix PAM system is the first supported system with more to come!

The external authentication system allows for specific users to be granted access to execute specific functions on specific minions. Access is configured in the master configuration file, and uses the new access control system:

external_auth:
  pam:
    thatch:
      - 'web*':
        - test.*
        - network.*

The configuration above allows the user thatch to execute functions in the test and network modules on minions that match the web* target.

Access Control System¶

All Salt systems can now be configured to grant access to non-administrative users in a granular way. The old configuration continues to work. Specific functions can be opened up to specific minions from specific users in the case of external auth and client ACLs, and for specific minions in the case of the peer system.

Access controls are configured like this:

client_acl:
  fred:
    - web\*:
      - pkg.list_pkgs
      - test.*
      - apache.*

Target by Network¶

A new matcher has been added to the system which allows for minions to be targeted by network. This new matcher can be called with the -S flag on the command line and is available in all places that the matcher system is available. Using it is simple:

$ salt -S '192.168.1.0/24' test.ping
$ salt -S '192.168.1.100' test.ping

Nodegroup Nesting¶

Previously a nodegroup was limited by not being able to include another nodegroup, this restraint has been lifted and now nodegroups will be expanded within other nodegroups with the N@ classifier.

Salt Key Delete by Glob¶

The ability to delete minion keys by glob has been added to salt-key. To delete all minion keys whose minion name starts with 'web':

$ salt-key -d 'web*'

Master Tops System¶

The external_nodes system has been upgraded to allow for modular subsystems to be used to generate the top file data for a highstate run.

The external_nodes option still works but will be deprecated in the future in favor of the new master_tops option.

Example of using master_tops:

master_tops:
  ext_nodes: cobbler-external-nodes

Next Level Solaris Support¶

A lot of work has been put into improved Solaris support by Romeo Theriault. Packaging modules (pkgadd/pkgrm and pkgutil) and states, cron support and user and group management have all been added and improved upon. These additions along with SMF (Service Management Facility) service support and improved Solaris grain detection in 0.10.3 add up to Salt becoming a great tool to manage Solaris servers with.

Security¶

A vulnerability in the security handshake was found and has been repaired, old minions should be able to connect to a new master, so as usual, the master should be updated first and then the minions.

Pillar Updates¶

The pillar communication has been updated to add some extra levels of verification so that the intended minion is the only one allowed to gather the data. Once all minions and the master are updated to salt 0.10.4 please activate pillar 2 by changing the pillar_version in the master config to 2. This will be set to 2 by default in a future release.

Salt 0.10.5 Release Notes¶

Salt 0.10.5 is ready, and comes with some great new features. A few more interfaces have been modularized, like the outputter system. The job cache system has been made more powerful and can now store and retrieve jobs archived in external databases. The returner system has been extended to allow minions to easily retrieve data from a returner interface.

As usual, this is an exciting release, with many noteworthy additions!

Major Features¶

External Job Cache¶

The external job cache is a system which allows for a returner interface to also act as a job cache. This system is intended to allow users to store job information in a central location for longer periods of time and to make the act of looking up information from jobs executed on other minions easier.

Currently the external job cache is supported via the mongo and redis returners:

ext_job_cache: redis
redis.host: salt

Once the external job cache is turned on the new ret module can be used on the minions to retrieve return information from the job cache. This can be a great way for minions to respond and react to other minions.

OpenStack Additions¶

OpenStack integration with Salt has been moving forward at a blistering pace. The new nova, glance and keystone modules represent the beginning of ongoing OpenStack integration.

The Salt team has had many conversations with core OpenStack developers and is working on linking to OpenStack in powerful new ways.

Wheel System¶

A new API was added to the Salt Master which allows the master to be managed via an external API. This new system allows Salt API to easily hook into the Salt Master and manage configs, modify the state tree, manage the pillar and more. The main motivation for the wheel system is to enable features needed in the upcoming web UI so users can manage the master just as easily as they manage minions.

The wheel system has also been hooked into the external auth system. This allows specific users to have granular access to manage components of the Salt Master.

Render Pipes¶

Jack Kuan has added a substantial new feature. The render pipes system allows Salt to treat the render system like unix pipes. This new system enables sls files to be passed through specific render engines. While the default renderer is still recommended, different engines can now be more easily merged. So to pipe the output of Mako used in YAML use this shebang line:

#!mako|yaml

Salt Key Overhaul¶

The Salt Key system was originally developed as only a CLI interface, but as time went on it was pressed into becoming a clumsy API. This release marks a complete overhaul of Salt Key. Salt Key has been rewritten to function purely from an API and to use the outputter system. The benefit here is that the outputter system works much more cleanly with Salt Key now, and the internals of Salt Key can be used much more cleanly.

Modular Outputters¶

The outputter system is now loaded in a modular way. This means that output systems can be more easily added by dropping a python file down on the master that contains the function output.

Gzip from Fileserver¶

Gzip compression has been added as an option to the cp.get_file and cp.get_dir commands. This will make file transfers more efficient and faster, especially over slower network links.

Unified Module Configuration¶

In past releases of Salt, the minions needed to be configured for certain modules to function. This was difficult because it required pre-configuring the minions. 0.10.5 changes this by making all module configs on minions search the master config file for values.

Now if a single database server is needed, then it can be defined in the master config and all minions will become aware of the configuration value.

Salt Call Enhancements¶

The salt-call command has been updated in a few ways. Now, salt-call can take the --return option to send the data to a returner. Also, salt-call now reports executions in the minion proc system, this allows the master to be aware of the operation salt-call is running.

Death to pub_refresh and sub_timeout¶

The old configuration values pub_refresh and sub_timeout have been removed. These options were in place to alleviate problems found in earlier versions of ZeroMQ which have since been fixed. The continued use of these options has proven to cause problems with message passing and have been completely removed.

Git Revision Versions¶

When running Salt directly from git (for testing or development, of course) it has been difficult to know exactly what code is being executed. The new versioning system will detect the git revision when building and how many commits have been made since the last release. A release from git will look like this:

0.10.4-736-gec74d69

Svn Module Addition¶

Anthony Cornehl (twinshadow) contributed a module that adds Subversion support to Salt. This great addition helps round out Salt's VCS support.

Noteworthy Changes¶

Arch Linux Defaults to Systemd¶

Arch Linux recently changed to use systemd by default and discontinued support for init scripts. Salt has followed suit and defaults to systemd now for managing services in Arch.

Salt, Salt Cloud and Openstack¶

With the releases of Salt 0.10.5 and Salt Cloud 0.8.2, OpenStack becomes the first (non-OS) piece of software to include support both on the user level (with Salt Cloud) and the admin level (with Salt). We are excited to continue to extend support of other platforms at this level.

Salt 0.11.0 Release Notes¶

Salt 0.11.0 is here, with some highly sought after and exciting features. These features include the new overstate system, the reactor system, a new state run scope component called __context__, the beginning of the search system (still needs a great deal of work), multiple package states, the MySQL returner and a better system to arbitrarily reference outputters.

It is also noteworthy that we are changing how we mark release numbers. For the life of the project we have been pushing every release with features and fixes as point releases. We will now be releasing point releases for only bug fixes on a more regular basis and major feature releases on a slightly less regular basis. This means that the next release will be a bugfix only release with a version number of 0.11.1. The next feature release will be named 0.12.0 and will mark the end of life for the 0.11 series.

Major Features¶

OverState¶

The overstate system is a simple way to manage rolling state executions across many minions. The overstate allows for a state to depend on the successful completion of another state.

Reactor System¶

The new reactor system allows for a reactive logic engine to be created which can respond to events within a salted environment. The reactor system uses sls files to match events fired on the master with actions, enabling Salt to react to problems in an infrastructure.

Your load-balanced group of webservers is under extra load? Spin up a new VM and add it to the group. Your fileserver is filling up? Send a notification to your sysadmin on call. The possibilities are endless!

Module Context¶

A new component has been added to the module loader system. The module context is a data structure that can hold objects for a given scope within the module.

This allows for components that are initialized to be stored in a persistent context which can greatly speed up ongoing connections. Right now the best example can be found in the cp execution module.

Multiple Package Management¶

A long desired feature has been added to package management. By definition Salt States have always installed packages one at a time. On most platforms this is not the fastest way to install packages. Erik Johnson, aka terminalmage, has modified the package modules for many providers and added new capabilities to install groups of packages. These package groups can be defined as a list of packages available in repository servers:

python_pkgs:
  pkg.installed:
    - pkgs:
      - python-mako
      - whoosh
      - python-git

or specify based on the location of specific packages:

python_pkgs:
  pkg.installed:
    - sources:
      - python-mako: http://some-rpms.org/python-mako.rpm
      - whoosh: salt://whoosh/whoosh.rpm
      - python-git: ftp://companyserver.net/python-git.rpm

Search System¶

The bones to the search system have been added. This is a very basic interface that allows for search backends to be added as search modules. The first supported search module is the whoosh search backend. Right now only the basic paths for the search system are in place, making this very experimental. Further development will involve improving the search routines and index routines for whoosh and other search backends.

The search system has been made to allow for searching through all of the state and pillar files, configuration files and all return data from minion executions.

Notable Changes¶

All previous versions of Salt have shared many directories between the master and minion. The default locations for keys, cached data and sockets has been shared by master and minion. This has created serious problems with running a master and a minion on the same systems. 0.11.0 changes the defaults to be separate directories. Salt will also attempt to migrate all of the old key data into the correct new directories, but if it is not successful it may need to be done manually. If your keys exhibit issues after updating make sure that they have been moved from /etc/salt/pki to /etc/salt/pki/{master,minion}.

The old setup will look like this:

/etc/salt/pki
|-- master.pem
|-- master.pub
|-- minions
|   `-- ragnarok.saltstack.net
|-- minions_pre
|-- minion.pem
|-- minion.pub
|-- minion_master.pub
|-- minions_pre
`-- minions_rejected

With the accepted minion keys in /etc/salt/pki/minions, the new setup places the accepted minion keys in /etc/salt/pki/master/minions.

/etc/salt/pki
|-- master
|   |-- master.pem
|   |-- master.pub
|   |-- minions
|   |   `-- ragnarok.saltstack.net
|   |-- minions_pre
|   `-- minions_rejected
|-- minion
|   |-- minion.pem
|   |-- minion.pub
|   `-- minion_master.pub

Salt 0.12.0 Release Notes¶

Another feature release of Salt is here! Some exciting additions are included with more ways to make salt modular and even easier management of the salt file server.

Major Features¶

Modular Fileserver Backend¶

The new modular fileserver backend allows for any external system to be used as a salt file server. The main benefit here is that it is now possible to tell the master to directly use a git remote location, or many git remote locations, automatically mapping git branches and tags to salt environments.

Windows is First Class!¶

A new Salt Windows installer is now available! Much work has been put in to improve Windows support. With this much easier method of getting Salt on your Windows machines, we hope even more development and progress will occur. Please file bug reports on the Salt GitHub repo issue tracker so we can continue improving.

One thing that is missing on Windows that Salt uses extensively is a software package manager and a software package repository. The Salt pkg state allows sys admins to install software across their infrastructure and across operating systems. Software on Windows can now be managed in the same way. The SaltStack team built a package manager that interfaces with the standard Salt pkg module to allow for installing and removing software on Windows. In addition, a software package repository has been built on top of the Salt fileserver. A small YAML file provides the information necessary for the package manager to install and remove software.

An interesting feature of the new Salt Windows software package repository is that one or more remote git repositories can supplement the master's local repository. The repository can point to software on the master's fileserver or on an HTTP, HTTPS, or ftp server.

New Default Outputter¶

Salt displays data to the terminal via the outputter system. For a long time the default outputter for Salt has been the python pretty print library. While this has been a generally reasonable outputter, it did have many failings. The new default outputter is called "nested", it recursively scans return data structures and prints them out cleanly.

If the result of the new nested outputter is not desired any other outputter can be used via the --out option, or the output option can be set in the master and minion configs to change the default outputter.

Internal Scheduler¶

The internal Salt scheduler is a new capability which allows for functions to be executed at given intervals on the minion, and for runners to be executed at given intervals on the master. The scheduler allows for sequences such as executing state runs (locally on the minion or remotely via an overstate) or continually gathering system data to be run at given intervals.

The configuration is simple, add the schedule option to the master or minion config and specify jobs to run, this in the master config will execute the state.over runner every 60 minutes:

schedule:
  overstate:
    function: state.over
    minutes: 60

This example for the minion configuration will execute a highstate every 30 minutes:

schedule:
  highstate:
    function: state.highstate
    minutes: 30

Optional DSL for SLS Formulas¶

Jack Kuan, our renderer expert, has created something that is astonishing. Salt, now comes with an optional Python based DSL, this is a very powerful interface that makes writing SLS files in pure python easier than it was with the raw py renderer. As usual this can be used with the renderer shebang line, so a single sls can be written with the DSL if pure python power is needed while keeping other sls files simple with YAML.

Set Grains Remotely¶

A new execution function and state module have been added that allows for grains to be set on the minion. Now grains can be set via a remote execution or via states. Use the grains.present state or the grains.setval execution functions.

Gentoo Additions¶

Major additions to Gentoo specific components have been made. The encompasses executions modules and states ranging from supporting the make.conf file to tools like layman.

Salt 0.13.0 Release Notes¶

The lucky number 13 has turned the corner! From CLI notifications when quitting a salt command, to substantial improvements on Windows, Salt 0.13.0 has arrived!

Major Features¶

Improved file.recurse Performance¶

The file.recurse system has been deployed and used in a vast array of situations. Fixes to the file state and module have led towards opening up new ways of running file.recurse to make it faster. Now the file.recurse state will download fewer files and will run substantially faster.

Windows Improvements¶

Minion stability on Windows has improved. Many file operations, including file.recurse, have been fixed and improved. The network module works better, to include network.interfaces. Both 32bit and 64bit installers are now available.

Nodegroup Targeting in Peer System¶

In the past, nodegroups were not available for targeting via the peer system. This has been fixed, allowing the new nodegroup expr_form argument for the publish.publish function:

salt-call publish.publish group1 test.ping expr_form=nodegroup

Blacklist Additions¶

Additions allowing more granular blacklisting are available in 0.13.0. The ability to blacklist users and functions in client_acl have been added, as well as the ability to exclude state formulas from the command line.

Command Line Pillar Embedding¶

Pillar data can now be embedded on the command line when calling state.sls and state.highstate. This allows for on the fly changes or settings to pillar and makes parameterizing state formulas even easier. This is done via the keyword argument:

salt '*' state.highstate pillar='{"cheese": "spam"}'

The above example will extend the existing pillar to hold the cheese key with a value of spam. If the cheese key is already specified in the minion's pillar then it will be overwritten.

CLI Notifications¶

In the past hitting ctrl-C and quitting from the salt command would just drop to a shell prompt, this caused confusion with users who expected the remote executions to also quit. Now a message is displayed showing what command can be used to track the execution and what the job id is for the execution.

Version Specification in Multiple-Package States¶

Versions can now be specified within multiple-package pkg.installed states. An example can be found below:

mypkgs:
  pkg.installed:
    - pkgs:
      - foo
      - bar: 1.2.3-4
      - baz

Noteworthy Changes¶

The configuration subsystem in Salt has been overhauled to make the opts dict used by Salt applications more portable, the problem is that this is an incompatible change with salt-cloud, and salt-cloud will need to be updated to the latest git to work with Salt 0.13.0. Salt Cloud 0.8.5 will also require Salt 0.13.0 or later to function.

The Salt Stack team is sorry for the inconvenience here, we work hard to make sure these sorts of things do not happen, but sometimes hard changes get in.

Salt 0.14.0 Release Notes¶

Salt 0.14.0 is here! This release was held up primarily by PyCon, Scale and illness, but has arrived! 0.14.0 comes with many new features and is breaking ground for Salt in the area of cloud management with the introduction of Salt providing basic cloud controller functionality.

Major Features¶

Salt - As a Cloud Controller¶

This is the first primitive inroad to using Salt as a cloud controller is available in 0.14.0. Be advised that this is alpha, only tested in a few very small environments.

The cloud controller is built using kvm and libvirt for the hypervisors. Hypervisors are autodetected as minions and only need to have libvirt running and kvm installed to function. The features of the Salt cloud controller are as follows:

Basic vm discovery and reporting

Creation of new virtual machines

Seeding virtual machines with Salt via qemu-nbd or libguestfs

Live migration (shared and non shared storage)

Delete existing VMs

It is noteworthy that this feature is still Alpha, meaning that all rights are reserved to change the interface if needs be in future releases!

Libvirt State¶

One of the problems with libvirt is management of certificates needed for live migration and cross communication between hypervisors. The new libvirt state makes the Salt Master hold a CA and manage the signing and distribution of keys onto hypervisors, just add a call to the libvirt state in the sls formulas used to set up a hypervisor:

libvirt_keys:
  libvirt.keys

New get Functions¶

An easier way to manage data has been introduced. The pillar, grains and config execution modules have been extended with the new get function. This function works much in the same way as the get method in a python dict, but with an enhancement, nested dict components can be extracted using a : delimiter.

If a structure like this is in pillar:

foo:
  bar:
    baz: quo

Extracting it from the raw pillar in an sls formula or file template is done this way:

{{ pillar['foo']['bar']['baz'] }}

Now with the new get function the data can be safely gathered and a default can be set allowing the template to fall back if the value is not available:

{{ salt['pillar.get']('foo:bar:baz', 'qux') }}

This makes handling nested structures much easier, and defaults can be cleanly set. This new function is being used extensively in the new formulae repository of salt sls formulas.

Salt 0.15.0 Release Notes¶

The many new features of Salt 0.15.0 have arrived! Salt 0.15.0 comes with many smaller features and a few larger ones.

These features range from better debugging tools to the new Salt Mine system.

Major Features¶

The Salt Mine¶

First there was the peer system, allowing for commands to be executed from a minion to other minions to gather data live. Then there was the external job cache for storing and accessing long term data. Now the middle ground is being filled in with the Salt Mine. The Salt Mine is a system used to execute functions on a regular basis on minions and then store only the most recent data from the functions on the master, then the data is looked up via targets.

The mine caches data that is public to all minions, so when a minion posts data to the mine all other minions can see it.

IPV6 Support¶

0.13.0 saw the addition of initial IPV6 support but errors were encountered and it needed to be stripped out. This time the code covers more cases and must be explicitly enabled. But the support is much more extensive than before.

Copy Files From Minions to the Master¶

Minions have long been able to copy files down from the master file server, but until now files could not be easily copied from the minion up to the master.

A new function called cp.push can push files from the minions up to the master server. The uploaded files are then cached on the master in the master cachedir for each minion.

Better Template Debugging¶

Template errors have long been a burden when writing states and pillar. 0.15.0 will now send the compiled template data to the debug log, this makes tracking down the intermittent stage templates much easier. So running state.sls or state.highstate with -l debug will now print out the rendered templates in the debug information.

State Event Firing¶

The state system is now more closely tied to the master's event bus. Now when a state fails the failure will be fired on the master event bus so that the reactor can respond to it.

Major Syndic Updates¶

The Syndic system has been basically re-written. Now it runs in a completely asynchronous way and functions primarily as an event broker. This means that the events fired on the syndic are now pushed up to the higher level master instead of the old method used which waited for the client libraries to return.

This makes the syndic much more accurate and powerful, it also means that all events fired on the syndic master make it up the pipe as well making a reactor on the higher level master able to react to minions further downstream.

Peer System Updates¶

The Peer System has been updated to run using the client libraries instead of firing directly over the publish bus. This makes the peer system much more consistent and reliable.

Minion Key Revocation¶

In the past when a minion was decommissioned the key needed to be manually deleted on the master, but now a function on the minion can be used to revoke the calling minion's key:

$ salt-call saltutil.revoke_auth

Function Return Codes¶

Functions can now be assigned numeric return codes to determine if the function executed successfully. While not all functions have been given return codes, many have and it is an ongoing effort to fill out all functions that might return a non-zero return code.

Functions in Overstate¶

The overstate system was originally created to just manage the execution of states, but with the addition of return codes to functions, requisite logic can now be used with respect to the overstate. This means that an overstate stage can now run single functions instead of just state executions.

Pillar Error Reporting¶

Previously if errors surfaced in pillar, then the pillar would consist of only an empty dict. Now all data that was successfully rendered stays in pillar and the render error is also made available. If errors are found in the pillar, states will refuse to run.

Using Cached State Data¶

Sometimes states are executed purely to maintain a specific state rather than to update states with new configs. This is grounds for the new cached state system. By adding cache=True to a state call the state will not be generated fresh from the master but the last state data to be generated will be used. If no previous state data is available then fresh data will be generated.

Monitoring States¶

The new monitoring states system has been started. This is very young but allows for states to be used to configure monitoring routines. So far only one monitoring state is available, the disk.status state. As more capabilities are added to Salt UI the monitoring capabilities of Salt will continue to be expanded.

Salt 0.15.1 Release Notes¶

The 0.15.1 release has been posted, this release includes fixes to a number of bugs in 0.15.1 and a three security patches.

Security Updates¶

A number of security issues have been resolved via the 0.15.1 release.

Path Injection in Minion IDs¶

Salt masters did not properly validate the id of a connecting minion. This can lead to an attacker uploading files to the master in arbitrary locations. In particular this can be used to bypass the manual validation of new unknown minions. Exploiting this vulnerability does not require authentication.

This issue affects all known versions of Salt.

This issue was reported by Ronald Volgers.

Patch¶

The issue is fixed in Salt 0.15.1. Updated packages are available in the usual locations.

Specific commits:

https://github.com/saltstack/salt/commit/5427b9438e452a5a8910d9128c6aafb45d8fd5d3

https://github.com/saltstack/salt/commit/7560908ee62351769c3cd43b03d74c1ca772cc52

https://github.com/saltstack/salt/commit/e200b8a7ff53780124e08d2bdefde7587e52bfca

RSA Key Generation Fault¶

RSA key generation was done incorrectly, leading to very insecure keys. It is recommended to regenerate all RSA keys.

This issue can be used to impersonate Salt masters or minions, or decrypt any transferred data.

This issue can only be exploited by attackers who are able to observe or modify traffic between Salt minions and the legitimate Salt master.

A tool was included in 0.15.1 to assist in mass key regeneration, the manage.regen_keys runner.

This issue affects all known versions of Salt.

This issue was reported by Ronald Volgers.

Patch¶

The issue is fixed in Salt 0.15.1. Updated packages are available in the usual locations.

Specific commits:

https://github.com/saltstack/salt/commit/5dd304276ba5745ec21fc1e6686a0b28da29e6fc

Command Injection Via ext_pillar¶

Arbitrary shell commands could be executed on the master by an authenticated minion through options passed when requesting a pillar.

Ext pillar options have been restricted to only allow safe external pillars to be called when prompted by the minion.

This issue affects Salt versions from 0.14.0 to 0.15.0.

This issue was reported by Ronald Volgers.

Patch¶

The issue is fixed in Salt 0.15.1. Updated packages are available in the usual locations.

Specific commits:

https://github.com/saltstack/salt/commit/43d8c16bd26159d827d1a945c83ac28159ec5865

Salt 0.16.0 Release Notes¶

The 0.16.0 release is an exciting one, with new features in master redundancy, and a new, powerful requisite.

Major Features¶

Multi-Master¶

This new capability allows for a minion to be actively connected to multiple salt masters at the same time. This allows for multiple masters to send out commands to minions and for minions to automatically reconnect to masters that have gone down. A tutorial is available to help get started here:

Multi Master Tutorial

Prereq, the New Requisite¶

The new prereq requisite is very powerful! It allows for states to execute based on a state that is expected to make changes in the future. This allows for a change on the system to be preempted by another execution. A good example is needing to shut down a service before modifying files associated with it, allowing, for instance, a webserver to be shut down allowing a load balancer to stop sending requests while server side code is updated. In this case, the prereq will only run if changes are expected to happen in the prerequired state, and the prerequired state will always run after the prereq state and only if the prereq state succeeds.

Peer System Improvements¶

The peer system has been revamped to make it more reliable, faster, and like the rest of Salt, async. The peer calls when an updated minion and master are used together will be much faster!

Relative Includes¶

The ability to include an sls relative to the defined sls has been added, the new syntax id documented here:

Includes

More State Output Options¶

The state_output option in the past only supported full and terse, 0.16.0 add the mixed and changes modes further refining how states are sent to users' eyes.

Improved Windows Support¶

Support for Salt on Windows continues to improve. Software management on Windows has become more seamless with Linux/UNIX/BSD software management. Installed software is now recognized by the short names defined in the repository SLS. This makes it possible to run salt '*' pkg.version firefox and get back results from Windows and non-Windows minions alike.

When templating files on Windows, Salt will now correctly use Windows appropriate line endings. This makes it much easier to edit and consume files on Windows.

When using the cmd state the shell option now allows for specifying Windows Powershell as an alternate shell to execute cmd.run and cmd.script. This opens up Salt to all the power of Windows Powershell and its advanced Windows management capabilities.

Several fixes and optimizations were added for the Windows networking modules, especially when working with IPv6.

A system module was added that makes it easy to restart and shutdown Windows minions.

The Salt Minion will now look for its config file in c:\salt\conf by default. This means that it's no longer necessary to specify the -c option to specify the location of the config file when starting the Salt Minion on Windows in a terminal.

Muliple Targets for pkg.removed, pkg.purged States¶

Both pkg.removed and pkg.purged now support the pkgs argument, which allow for multiple packages to be targeted in a single state. This, as in pkg.installed, helps speed up these states by reducing the number of times that the package management tools (apt, yum, etc.) need to be run.

Random Times in Cron States¶

The temporal parameters in cron.present states (minute, hour, etc.) can now be randomized by using random instead of a specific value. For example, by using the random keyword in the minute parameter of a cron state, the same cron job can be pushed to hundreds or thousands of hosts, and they would each use a randomly-generated minute. This can be helpful when the cron job accesses a network resource, and it is not desirable for all hosts to run the job concurrently.

/path/to/cron/script:
  cron.present:
    - user: root
    - minute: random
    - hour: 2

Since Salt assumes a value of * for unspecified temporal parameters, adding a parameter to the state and setting it to random will change that value from * to a randomized numeric value. However, if that field in the cron entry on the minion already contains a numeric value, then using the random keyword will not modify it.

Confirmation Prompt on Key Acceptance¶

When accepting new keys with salt-key -a minion-id or salt-key -A, there is now a prompt that will show the affected keys and ask for confirmation before proceeding. This prompt can be bypassed using the -y or --yes command line argument, as with other salt-key commands.

Support for Setting Password Hashes on BSD Minions¶

FreeBSD, NetBSD, and OpenBSD all now support setting passwords in user.present states.

Salt 0.16.2 Release Notes¶

Version 0.16.2 is a bugfix release for 0.16.0, and contains a number of fixes.

Windows¶

Only allow Administrator's group and SYSTEM user access to C:\salt. This eliminates a race condition where a non-admin user could modify a template or managed file before it is executed by the minion (which is running as an elevated user), thus avoiding a potential escalation of privileges. (issue 6361)

Grains¶

Fixed detection of virtual grain on OpenVZ hardware nodes
Gracefully handle lsb_release data when it is enclosed in quotes
LSB grains are now prefixed with lsb_distrib_ instead of simply lsb_. The old naming is not preserved, so SLS may be affected.
Improved grains detection on MacOS

Pillar¶

Don't try to load git_pillar if not enabled in master config (issue 6052)
Functions pillar.item and pillar.items added for parity with grains.item/grains.items. The old function pillar.data is preserved for backwards compatibility.
Fixed minion traceback when Pillar SLS is malformed (issue 5910)

Peer Publishing¶

More gracefully handle improperly quoted publish commands (issue 5958)
Fixed traceback when timeout specified via the CLI fo publish.publish, publish.full_data (issue 5959)
Fixed unintended change in output of publish.publish (issue 5928)

Minion¶

Fixed salt-key usage in minionswarm script
Quieted warning about SALT_MINION_CONFIG environment variable on minion startup and for CLI commands run via salt-call (issue 5956)
Added minion config parameter random_reauth_delay to stagger re-auth attempts when the minion is waiting for the master to approve its public key. This helps prevent SYN flooding in larger environments.

User/Group Management¶

Implement previously-ignored unique option for user.present states in FreeBSD
Report in state output when a group.present state attempts to use a gid in use by another group
Fixed regression that prevents a user.present state to set the password hash to the system default (i.e. an unset password)
Fixed multiple group.present states with the same group (issue 6439)

File Management¶

Fixed file.mkdir setting incorrect permissions (issue 6033)
Fixed cleanup of source files for templates when /tmp is in file_roots (issue 6118)
Fixed caching of zero-byte files when a non-empty file was previously cached at the same path
Added HTTP authentication support to the cp module (issue 5641)
Diffs are now suppressed when binary files are changed

Package/Repository Management¶

Fixed traceback when there is only one target for pkg.latest states
Fixed regression in detection of virtual packages (apt)
Limit number of pkg database refreshes to once per state.sls/state.highstate
YUM: Allow 32-bit packages with arches other than i686 to be managed on 64-bit systems (issue 6299)
Fixed incorrect reporting in pkgrepo.managed states (issue 5517)
Fixed 32-bit binary package installs on 64-bit RHEL-based distros, and added proper support for 32-bit packages on 64-bit Debian-based distros (issue 6303)
Fixed issue where requisites were inadvertently being put into YUM repo files (issue 6471)

Service Management¶

Fixed inaccurate reporting of results in service.running states when the service fails to start (issue 5894)
Fixed handling of custom initscripts in RHEL-based distros so that they are immediately available, negating the need for a second state run to manage the service that the initscript controls

Networking¶

Function network.hwaddr renamed to network.hw_addr to match network.ip_addrs and network.ip_addrs6. All three functions also now work without the underscore in the name, as well.
Fixed traceback in bridge.show when interface is not present (issue 6326)

SSH¶

Fixed incorrect result reporting for some ssh_known_hosts.present states
Fixed inaccurate reporting when ssh_auth.present states are run with test=True, when rsa/dss is used for the enc param instead of ssh-rsa/ssh-dss (issue 5374)

pip¶

Properly handle -f lines in pip freeze output
Fixed regression in pip.installed states with specifying a requirements file (issue 6003)
Fixed use of editable argument in pip.installed states (issue 6025)
Deprecated runas parameter in execution function calls, in favor of user

MySQL¶

Allow specification of MySQL connection arguments via the CLI, overriding/bypassing minion config params
Allow mysql_user.present states to set a passwordless login (issue 5550)
Fixed endless loop when mysql.processlist is run (issue 6297)

PostgreSQL¶

Fixed traceback in postgres.user_list (issue 6352)

Miscellaneous¶

Don't allow npm states to be used if npm module is not available
Fixed alternatives.install states for which the target is a symlink (issue 6162)
Fixed traceback in sysbench module (issue 6175)
Fixed traceback in job cache
Fixed tempfile cleanup for windows
Fixed issue where SLS files using the pydsl renderer were not being run
Fixed issue where returners were being passed incorrect information (issue 5518)
Fixed traceback when numeric args are passed to cmd.script states
Fixed bug causing cp.get_dir to return more directories than expected (issue 6048)
Fixed traceback when supervisord.running states are run with test=True (issue 6053)
Fixed tracebacks when Salt encounters problems running rbenv (issue 5888)
Only make the monit module available if monit binary is present (issue 5871)
Fixed incorrect behavior of img.mount_image
Fixed traceback in tomcat.deploy_war in Windows
Don't re-write /etc/fstab if mount fails
Fixed tracebacks when Salt encounters problems running gem (issue 5886)
Fixed incorrect behavior of selinux.boolean states (issue 5912)
RabbitMQ: Quote passwords to avoid symbols being interpolated by the shell (issue 6338)
Fixed tracebacks in extfs.mkfs and extfs.tune (issue 6462)
Fixed a regression with the module.run state where the m_name and m_fun arguments were being ignored (issue 6464)

Salt 0.16.3 Release Notes¶

Version 0.16.3 is another bugfix release for 0.16.0. The changes include:

Various documentation fixes
Fix proc directory regression (issue 6502)
Properly detect Linaro Linux (issue 6496)
Fix regressions in mount.mounted (issue 6522, issue 6545)
Skip malformed state requisites (issue 6521)
Fix regression in gitfs from bad import
Fix for watching prereq states (including recursive requisite error) (issue 6057)
Fix mod_watch not overriding prereq (issue 6520)
Don't allow functions which compile states to be called within states (issue 5623)
Return error for malformed top.sls (issue 6544)
Fix traceback in mysql.query
Fix regression in binary package installation for 64-bit packages on Debian-based Linux distros (issue 6563)
Fix traceback caused by running cp.push without having set file_recv in the master config file
Fix scheduler configuration in pillar (issue 6201)

Salt 0.16.4 Release Notes¶

Version 0.16.4 is another bugfix release for 0.16.0, likely to be the last before 0.17.0 is released. The changes include:

Multiple documentation improvements/additions
Added the osfinger and osarch grains
Properly handle 32-bit packages for debian32 on x86_64 (issue 6607)
Fix regression in yum package installation in CentOS 5 (issue 6677)
Fix bug in hg.latest state that would erroneously delete directories (issue 6661)
Fix bug related to pid not existing for ps.top (issue 6679)
Fix regression in MySQL returner (issue 6695)
Fix IP addresses grains (ipv4 and ipv6) to include all addresses (issue 6656)
Fix regression preventing authenticated FTP (issue 6733)
Fix setting password for windows users (issue 6824)
Fix file.contains on values YAML parses as non-string (issue 6817)
Fix file.get_gid, file.get_uid, and file.chown for broken symlinks (issue 6826)
Fix comment for service reloads in service state (issue 6851)

Salt 0.17.0 Release Notes¶

The 0.17.0 release is a very exciting release of Salt, this brings to Salt some very powerful new features and advances. The advances range from the state system to the test suite, covering new transport capabilities and making states easier and more powerful, to extending Salt Virt and much more!

The 0.17.0 release will also be the last release of Salt to follow the old 0.XX.X numbering system, the next release of Salt will change the numbering to be date based following this format:

So if the release happens in November of 2013 the number will be 13.11.0, the first bugfix release will be 13.11.1 and so forth.

Major Features¶

Halite¶

The new Halite web GUI is now available, a great deal of work has been put into Halite to make it fully event driven and amazingly fast. The Halite UI can be started from within the Salt Master, or standalone, and does not require an external database to run, it is very lightweight!

This initial release of Halite is primarily the framework for the UI and the communication systems making it easy to extend and build the UI up. It presently supports watching the event bus and firing commands over Salt.

Halite is, like the rest of Salt, Open Source!

Much more will be coming in the future of Halite!

Salt SSH¶

The new salt-ssh command has been added to Salt. This system allows for remote execution and states to be run over ssh. The benefit here being, that salt can run relying only on the ssh agent, rather than requiring a minion to be deployed.

The salt-ssh system runs states in a compatible way as Salt and states created and run with salt-ssh can be moved over to a standard salt deployment without modification.

Since this is the initial release of salt-ssh, there is plenty of room for improvement, but it is fully operational, not just a bootstrap tool.

Rosters¶

Salt is designed to have the minions be aware of the master and the master does not need to be aware of the location of the minions. The new salt roster system was created and designed to facilitate listing the targets for salt-ssh.

The roster system, like most of Salt, is a plugin system, allowing for the list of systems to target to be derived from any pluggable backend. The rosters shipping with 0.17.0 are flat and scan. Flat is a file which is read in via the salt render system and the scan roster does simple network scanning to discover ssh servers.

State Auto Order¶

This is a major change in how states are evaluated in Salt. State Auto Order is a new feature that makes states get evaluated and executed in the order in which they are defined in the sls file. This feature makes it very easy to see the finite order in which things will be executed, making Salt now, fully imperative AND fully declarative.

The requisite system still takes precedence over the order in which states are defined, so no existing states should break with this change. But this new feature can be turned off by setting state_auto_order: False in the master config, thus reverting to the old lexicographical order.

state.sls Runner¶

The state.sls runner has been created to allow for a more powerful system for orchestrating state runs and function calls across the salt minions. This new system uses the state system for organizing executions.

This allows for states to be defined that are executed on the master to call states on minions via salt-run state.sls.

Event Namespacing¶

Events have been updated to be much more flexible. The tags in events have all been namespaced allowing easier tracking of event names.

Mercurial Fileserver Backend¶

The popular git fileserver backend has been joined by the mercurial fileserver backend, allowing the state tree to be managed entirely via mercurial.

External Logging Handlers¶

The external logging handler system allows for Salt to directly hook into any external logging system. Currently supported are sentry and logstash.

Jenkins Testing¶

The testing systems in Salt have been greatly enhanced, tests for salt are now executed, via jenkins.saltstack.com, across many supported platforms. Jenkins calls out to salt-cloud to create virtual machines on Rackspace, then the minion on the virtual machine checks into the master running on Jenkins where a state run is executed that sets up the minion to run tests and executes the test suite.

This now automates the sequence of running platform tests and allows for continuous destructive tests to be run.

Salt Testing Project¶

The testing libraries for salt have been moved out of the main salt code base and into a standalone codebase. This has been done to ease the use of the testing systems being used in salt based projects other than Salt itself.

StormPath External Authentication¶

The external auth system now supports the fantastic Stormpath cloud based authentication system.

LXC Support¶

Extensive additions have been added to Salt for LXC support. This included the backend libs for managing LXC containers. Addition into the salt-virt system is still in the works.

Mac OS X User/Group Support¶

Salt is now able to manage users and groups on Minions running Mac OS X. However, at this time user passwords cannot be managed.

Django ORM External Pillar¶

Pillar data can now be derived from Django managed databases.

Fixes from RC to release¶

Multiple documentation fixes
Add multiple source files + templating for file.append (issue 6905)
Support sysctl configuration files in systemd>=207 (issue 7351)
Add file.search and file.replace
Fix cross-calling execution functions in provider overrides
Fix locale override for postgres (issue 4543)
Fix Raspbian identification for service/pkg support (issue 7371)
Fix cp.push file corruption (issue 6495)
Fix ALT Linux password hash specification (issue 3474)
Multiple salt-ssh-related fixes and improvements

Salt 0.6.0 release notes¶

The Salt remote execution manager has reached initial functionality! Salt is a management application which can be used to execute commands on remote sets of servers.

The whole idea behind Salt is to create a system where a group of servers can be remotely controlled from a single master, not only can commands be executed on remote systems, but salt can also be used to gather information about your server environment.

Unlike similar systems, like Func and MCollective, Salt is extremely simple to setup and use, the entire application is contained in a single package, and the master and minion daemons require no running dependencies in the way that Func requires Certmaster and MCollective requires activeMQ.

Salt also manages authentication and encryption. Rather than using SSL for encryption, salt manages encryption on a payload level, so the data sent across the network is encrypted with fast AES encryption, and authentication uses RSA keys. This means that Salt is fast, secure, and very efficient.

Messaging in Salt is executed with ZeroMQ, so the message passing interface is built into salt and does not require an external ZeroMQ server. This also adds speed to Salt since there is no additional bloat on the networking layer, and ZeroMQ has already proven itself as a very fast networking system.

The remote execution in Salt is "Lazy Execution", in that once the command is sent the requesting network connection is closed. This makes it easier to detach the execution from the calling process on the master, it also means that replies are cached, so that information gathered from historic commands can be queried in the future.

Salt also allows users to make execution modules in Python. Writers of these modules should also be pleased to know that they have access to the impressive information gathered from PuppetLabs' Facter application, making Salt module more flexible. In the future I hope to also allow Salt to group servers based on Facter information as well.

All in all Salt is fast, efficient and clean, can be used from a simple command line client or through an API, uses message queue technology to make network execution extremely fast, and encryption is handled in a very fast and efficient manner. Salt is also VERY easy to use and VERY easy to extend.

You can find the source code for Salt on my GitHub page, I have also set up a few wiki pages explaining how to use and set up Salt. If you are using Arch Linux there is a package available in the Arch Linux AUR.

Salt 0.6.0 Source: https://github.com/downloads/saltstack/salt/salt-0.6.0.tar.gz

GitHub page: https://github.com/saltstack/salt

Wiki: https://github.com/saltstack/salt/wiki

Arch Linux Package: https://aur.archlinux.org/packages.php?ID=47512

I am very open to contributions, for instance I need packages for more Linux distributions as well as BSD packages and testers.

Give Salt a try, this is the initial release and is not a 1.0 quality release, but it has been working well for me! I am eager to get your feedback!

Salt 0.7.0 release notes¶

I am pleased to announce the release of Salt 0.7.0!

This release marks what is the first stable release of salt, 0.7.0 should be suitable for general use.

0.7.0 Brings the following new features to Salt:

Integration with Facter data from puppet labs
Allow for matching minions from the salt client via Facter information
Minion job threading, many jobs can be executed from the master at once
Preview of master clustering support - Still experimental
Introduce new minion modules for stats, virtualization, service management and more
Add extensive logging to the master and minion daemons
Add sys.reload_functions for dynamic function reloading
Greatly improve authentication
Introduce the saltkey command for managing public keys
Begin backend development preparatory to introducing butter
Addition of man pages for the core commands
Extended and cleaned configuration

0.7.0 Fixes the following major bugs:

Fix crash in minions when matching failed
Fix configuration file lookups for the local client
Repair communication bugs in encryption
Numerous fixes in the minion modules

The next release of Salt should see the following features:

Stabilize the cluster support
Introduce a remote client for salt command tiers
salt-ftp system for distributed file copies
Initial support for "butter"

Coming up next is a higher level management framework for salt called Butter. I want salt to stay as a simple and effective communication framework, and allow for more complicated executions to be managed via Butter.

Right now Butter is being developed to act as a cloud controller using salt as the communication layer, but features like system monitoring and advanced configuration control (a puppet manager) are also in the pipe.

Special thanks to Joseph Hall for the status and network modules, and thanks to Matthias Teege for tracking down some configuration bugs!

Salt can be downloaded from the following locations;

Source Tarball:

https://github.com/downloads/saltstack/salt/salt-0.7.0.tar.gz

Arch Linux Package:

https://aur.archlinux.org/packages.php?ID=47512

Please enjoy the latest Salt release!

Salt 0.8.0 release notes¶

Salt 0.8.0 is ready for general consumption! The source tarball is available on GitHub for download:

https://github.com/downloads/saltstack/salt/salt-0.8.0.tar.gz

A lot of work has gone into salt since the last release just 2 weeks ago, and salt has improved a great deal. A swath of new features are here along with performance and threading improvements!

The main new features of salt 0.8.0 are:

Salt-cp

Cython minion modules

Dynamic returners

Faster return handling

Lowered required Python version to 2.6

Advanced minion threading

Configurable minion modules

Salt-cp -¶

The salt-cp command introduces the ability to copy simple files via salt to targeted servers. Using salt-cp is very simple, just call salt-cp with a target specification, the source file(s) and where to copy the files on the minions. For instance:

# salt-cp ‘*’ /etc/hosts /etc/hosts

Will copy the local /etc/hosts file to all of the minions.

Salt-cp is very young, in the future more advanced features will be added, and the functionality will much more closely resemble the cp command.

Cython minion modules -¶

Cython is an amazing tool used to compile Python modules down to c. This is arguably the fastest way to run Python code, and since pyzmq requires cython, adding support to salt for cython adds no new dependencies.

Cython minion modules allow minion modules to be written in cython and therefore executed in compiled c. Simply write the salt module in cython and use the file extension “.pyx” and the minion module will be compiled when the minion is started. An example cython module is included in the main distribution called cytest.pyx:

https://github.com/saltstack/salt/blob/develop/salt/modules/cytest.pyx

Dynamic Returners -¶

By default salt returns command data back to the salt master, but now salt can return command data to any system. This is enabled via the new returners modules feature for salt. The returners modules take the return data and sends it to a specific module. The returner modules work like minion modules, so any returner can be added to the minions.

This means that a custom data returner can be added to communicate the return data so anything from MySQL, Redis, MongoDB and more!

There are 2 simple stock returners in the returners directory:

https://github.com/saltstack/salt/blob/develop/salt/returners

The documentation on writing returners will be added to the wiki shortly, and returners can be written in pure Python, or in cython.

Configurable Minion Modules -¶

Minion modules may need to be configured, now the options passed to the minion configuration file can be accessed inside of the minion modules via the __opt__ dict.

Information on how to use this simple addition has been added to the wiki: https://github.com/thatch45/salt/wiki/Writing-Salt-Modules

The test module has an example of using the __opts__ dict, and how to set default options:

https://github.com/saltstack/salt/blob/develop/salt/modules/test.py

Advanced Minion Threading:¶

In 0.7.0 the minion would block after receiving a command from the master, now the minion will spawn a thread or multiprocess. By default Python threads are used because for general use they have proved to be faster, but the minion can now be configured to use the Python multiprocessing module instead. Using multiprocessing will cause executions that are CPU bound or would otherwise exploit the negative aspects of the Python GIL to run faster and more reliably, but simple calls will still be faster with Python threading. The configuration option can be found in the minion configuration file:

https://github.com/saltstack/salt/blob/develop/conf/minion

Lowered Supported Python to 2.6 -

The requirement for Python 2.7 has been removed to support Python 2.6. I have received requests to take the minimum Python version back to 2.4, but unfortunately this will not be possible, since the ZeroMQ Python bindings do not support Python 2.4.

Salt 0.8.0 is a very major update, it also changes the network protocol slightly which makes communication with older salt daemons impossible, your master and minions need to be upgraded together! I could use some help bringing salt to the people! Right now I only have packages for Arch Linux, Fedora 14 and Gentoo. We need packages for Debian and people willing to help test on more platforms. We also need help writing more minion modules and returner modules. If you want to contribute to salt please hop on the mailing list and send in patches, make a fork on GitHub and send in pull requests! If you want to help but are not sure where you can, please email me directly or post tot he mailing list!

I hope you enjoy salt, while it is not yet 1.0 salt is completely viable and usable!

-Thomas S. Hatch

Salt 0.8.7 release notes¶

It has been a month since salt 0.8.0, and it has been a long month! But Salt is still coming along strong. 0.8.7 has a lot of changes and a lot of updates. This update makes Salt’s ZeroMQ back end better, strips Facter from the dependencies, and introduces interfaces to handle more capabilities.

Many of the major updates are in the background, but the changes should shine through to the surface. A number of the new features are still a little thin, but the back end to support expansion is in place.

I also recently gave a presentation to the Utah Python users group in Salt Lake City, the slides from this presentation are available here: https://github.com/downloads/saltstack/salt/Salt.pdf

The video from this presentation will be available shortly.

The major new features and changes in Salt 0.8.7 are:

Revamp ZeroMQ topology on the master for better scalability
State enforcement
Dynamic state enforcement managers
Extract the module loader into salt.loader
Make Job ids more granular
Replace Facter functionality with the new salt grains interface
Support for “virtual” salt modules
Introduce the salt-call command
Better debugging for minion modules

The new ZeroMQ topology allows for better scalability, this will be required by the need to execute massive file transfers to multiple machines in parallel and state management. The new ZeroMQ topology is available in the aforementioned presentation.

0.8.7 introduces the capability to declare states, this is similar to the capabilities of Puppet. States in salt are declared via state data structures. This system is very young, but the core feature set is available. Salt states work around rendering files which represent Salt high data. More on the Salt state system will be documented in the near future.

The system for loading salt modules has been pulled out of the minion class to be a standalone module, this has enabled more dynamic loading of Salt modules and enables many of the updates in 0.8.7 –

https://github.com/saltstack/salt/blob/develop/salt/loader.py

Salt Job ids are now microsecond precise, this was needed to repair a race condition unveiled by the speed improvements in the new ZeroMQ topology.

The new grains interface replaces the functionality of Facter, the idea behind grains differs from Facter in that the grains are only used for static system data, dynamic data needs to be derived from a call to a salt module. This makes grains much faster to use, since the grains data is generated when the minion starts.

Virtual salt modules allows for a salt module to be presented as something other than its module name. The idea here is that based on information from the minion decisions about which module should be presented can be made. The best example is the pacman module. The pacman module will only load on Arch Linux minions, and will be called pkg. Similarly the yum module will be presented as pkg when the minion starts on a Fedora/RedHat system.

The new salt-call command allows for minion modules to be executed from the minion. This means that on the minion a salt module can be executed, this is a great tool for testing Salt modules. The salt-call command can also be used to view the grains data.

In previous releases when a minion module threw an exception very little data was returned to the master. Now the stack trace from the failure is returned making debugging of minion modules MUCH easier.

Salt is nearing the goal of 1.0, where the core feature set and capability is complete!

Salt 0.8.7 can be downloaded from GitHub here: https://github.com/downloads/saltstack/salt/salt-0.8.7.tar.gz

-Thomas S Hatch

Salt 0.8.8 release notes¶

Salt 0.8.8 is here! This release adds a great deal of code and some serious new features. The latest release can be downloaded here: https://github.com/downloads/saltstack/salt/salt-0.8.8.tar.gz

Improved Documentation has been set up for salt using sphinx thanks to the efforts of Seth House. This new documentation system will act as the back end to the salt website which is still under heavy development. The new sphinx documentation system has also been used to greatly clean up the salt manpages. The salt 7 manpage in particular now contains extensive information which was previously only in the wiki. The new documentation can be found at: http://thatch45.github.com/salt-www/ We still have a lot to add, and when the domain is set up I will post another announcement.

More additions have been made to the ZeroMQ setup, particularly in the realm of file transfers. Salt 0.8.8 introduces a built in, stateless, encrypted file server which allows salt minions to download files from the salt master using the same encryption system used for all other salt communications. The main motivation for the salt file server has been to facilitate the new salt state system.

Much of the salt code has been cleaned up and a new cleaner logging system has been introduced thanks to the efforts of Pedro Algarvio. These additions will allow for much more flexible logging to be executed by salt, and fixed a great deal of my poor spelling in the salt docstrings! Pedro Algarvio has also cleaned up the API, making it easier to embed salt into another application.

The biggest addition to salt found in 0.8.8 is the new state system. The salt module system has received a new front end which allows salt to be used as a configuration management system. The configuration management system allows for system configuration to be defined in data structures. The configuration management system, or as it is called in salt, the “salt state system” supports many of the features found in other configuration managers, but allows for system states to be written in a far simpler format, executes at blazing speeds, and operates via the salt minion matching system. The state system also operates within the normal scope of salt, and requires no additional configuration to use.

The salt state system can enforce the following states with many more to come: Packages Files Services Executing commands Hosts

The system used to define the salt states is based on a data structure, the data structure used to define the salt states has been made to be as easy to use as possible. The data structure is defined by default using a YAML file rendered via a Jinja template. This means that the state definition language supports all of the data structures that YAML supports, and all of the programming constructs and logic that Jinja supports. If the user does not like YAML or Jinja the states can be defined in yaml-mako, json-jinja, or json-mako. The system used to render the states is completely dynamic, and any rendering system can be added to the capabilities of Salt, this means that a rendering system that renders XML data in a cheetah template, or whatever you can imagine, can be easily added to the capabilities of salt.

The salt state system also supports isolated environments, as well as matching code from several environments to a single salt minion.

The feature base for Salt has grown quite a bit since my last serious documentation push. As we approach 0.9.0 the goals are becoming very clear, and the documentation needs a lot of work. The main goals for 0.9.0 are to further refine the state system, fix any bugs we find, get Salt running on as many platforms as we can, and get the documentation filled out. There is a lot more to come as Salt moves forward to encapsulate a much larger scope, while maintaining supreme usability and simplicity.

If you would like a more complete overview of Salt please watch the Salt presentation: Flash Video: http://blip.tv/thomas-s-hatch/salt-0-8-7-presentation-5180182 OGV Video Download: http://blip.tv/file/get/Thatch45-Salt087Presentation416.ogv Slides: https://github.com/downloads/saltstack/salt/Salt.pdf

-Thomas S Hatch

Salt 0.8.9 Release Notes¶

Salt 0.8.9 has finally arrived! Unfortunately this is much later than I had hoped to release 0.8.9, life has been very crazy over the last month. But despite challenges, Salt has moved forward!

This release, as expected, adds few new features and many refinements. One of the most exciting aspect of this release is that the development community for salt has grown a great deal and much of the code is from contributors.

Also, I have filled out the documentation a great deal. So information on States is properly documented, and much of the documentation that was out of date has been filled in.

Download!¶

The Salt source can be downloaded from the salt GitHub site:

https://github.com/downloads/saltstack/salt/salt-0.8.9.tar.gz

Or from PyPI:

http://pypi.python.org/packages/source/s/salt/salt-0.8.9.tar.gz

Here s the md5sum:

7d5aca4633bc22f59045f59e82f43b56

For instructions on how to set up Salt please see the Installation instructions.

New Features¶

Salt Run¶

A big feature is the addition of Salt run, the salt-run command allows for master side execution modules to be made that gather specific information or execute custom routines from the master.

Documentation for salt-run can be found here:

http://saltstack.org/ref/runners.html

Refined Outputters¶

One problem often complained about in salt was the fact that the output was so messy. Thanks to help from Jeff Schroeder a cleaner interface for the command output for the Salt CLI has been made. This new interface makes adding new printout formats easy and additions to the capabilities of minion modules makes it possible to set the printout mode or outputter for functions in minion modules.

Cross Calling Salt Modules¶

Salt modules can now call each other, the __salt__ dict has been added to the predefined references in minion modules. This new feature is documented in the modules documentation:

http://saltstack.org/ref/modules/index.html

Watch Option Added to Salt State System¶

Now in Salt states you can set the watch option, this will allow watch enabled states to change based on a change in the other defined states. This is similar to subscribe and notify statements in puppet.

Root Dir Option¶

Travis Cline has added the ability to define the option root_dir which allows the salt minion to operate in a subdir. This is a strong move in supporting the minion running as an unprivileged user

Config Files Defined in Variables¶

Thanks again to Travis Cline, the master and minion configuration file locations can be defined in environment variables now.

New Modules¶

Quite a few new modules, states, returners and runners have been made.

New Minion Modules¶

apt¶

Support for apt-get has been added, this adds greatly improved Debian and Ubuntu support to Salt!

useradd and groupadd¶

Support for manipulating users and groups on Unix-like systems.

moosefs¶

Initial support for reporting on aspects of the distributed file system, MooseFS. For more information on MooseFS please see: http://moosefs.org

Thanks to Joseph Hall for his work on MooseFS support.

mount¶

Manage mounts and the fstab.

puppet¶

Execute puppet on remote systems.

shadow¶

Manipulate and manage the user password file.

ssh¶

Interact with ssh keys.

New States¶

user and group¶

Support for managing users and groups in Salt States.

mount¶

Enforce mounts and the fstab.

New Returners¶

mongo_return¶

Send the return information to a MongoDB server.

New Runners¶

manage¶

Display minions that are up or down.

Salt 0.9.0 Release Notes¶

Salt 0.9.0 is here. This is an exciting release, 0.9.0 includes the new network topology features allowing peer salt commands and masters of masters via the syndic interface.

0.9.0 also introduces many more modules, improvements to the API and improvements to the ZeroMQ systems.

Download!¶

The Salt source can be downloaded from the salt GitHub site:

https://github.com/downloads/saltstack/salt/salt-0.9.0.tar.gz

Or from PyPI:

http://pypi.python.org/packages/source/s/salt/salt-0.9.0.tar.gz

Here is the md5sum:

9a925da04981e65a0f237f2e77ddab37

For instructions on how to set up Salt please see the Installation instructions.

New Features¶

Salt Syndic¶

The new Syndic interface allows a master to be commanded via another higher level salt master. This is a powerful solution allowing a master control structure to exist, allowing salt to scale to much larger levels then before.

Peer Communication¶

0.9.0 introduces the capability for a minion to call a publication on the master and receive the return from another set of minions. This allows salt to act as a communication channel between minions and as a general infrastructure message bus.

Peer communication is turned off by default but can be enabled via the peer option in the master configuration file. Documentation on the new Peer interface.

Easily Extensible API¶

The minion and master classes have been redesigned to allow for specialized minion and master servers to be easily created. An example on how this is done for the master can be found in the master.py salt module:

https://github.com/saltstack/salt/blob/develop/salt/master.py

The Master class extends the SMaster class and set up the main master server.

The minion functions can now also be easily added to another application via the SMinion class, this class can be found in the minion.py module:

https://github.com/saltstack/salt/blob/develop/salt/minion.py

Cleaner Key Management¶

This release changes some of the key naming to allow for multiple master keys to be held based on the type of minion gathering the master key.

The -d option has also been added to the salt-key command allowing for easy removal of accepted public keys.

The --gen-keys option is now available as well for salt-key, this allows for a salt specific RSA key pair to be easily generated from the command line.

Improved 0MQ Master Workers¶

The 0MQ worker system has been further refined to be faster and more robust. This new system has been able to handle a much larger load than the previous setup. The new system uses the IPC protocol in 0MQ instead of TCP.

New Modules¶

Quite a few new modules have been made.

New Minion Modules¶

apache¶

Work directly with apache servers, great for managing balanced web servers

cron¶

Read out the contents of a systems crontabs

mdadm¶

Module to manage raid devices in Linux, appears as the raid module

mysql¶

Gather simple data from MySQL databases

ps¶

Extensive utilities for managing processes

publish¶

Used by the peer interface to allow minions to make publications

Salt 0.9.2 Release Notes¶

Salt 0.9.2 has arrived! 0.9.2 is primarily a bugfix release, the exciting component in 0.9.2 is greatly improved support for salt states. All of the salt states interfaces have been more thoroughly tested and the new salt-states git repo is growing with example of how to use states.

This release introduces salt states for early developers and testers to start helping us clean up the states interface and make it ready for the world!

0.9.2 also fixes a number of bugs found on Python 2.6.

Download!¶

The Salt source can be downloaded from the salt GitHub site:

https://github.com/downloads/saltstack/salt/salt-0.9.2.tar.gz

Or from PyPI:

http://pypi.python.org/packages/source/s/salt/salt-0.9.2.tar.gz

For instructions on how to set up Salt please see the Installation instructions.

New Features¶

Salt-Call Additions¶

The salt-call command has received an overhaul, it now hooks into the outputter system so command output looks clean, and the logging system has been hooked into salt-call, so the -l option allows the logging output from salt minion functions to be displayed.

The end result is that the salt-call command can execute the state system and return clean output:

# salt-call state.highstate

State System Fixes¶

The state system has been tested and better refined. As of this release the state system is ready for early testers to start playing with. If you are interested in working with the state system please check out the (still very small) salt-states GitHub repo:

https://github.com/thatch45/salt-states

This git repo is the active development branch for determining how a clean salt-state database should look and act. Since the salt state system is still very young a lot of help is still needed here. Please fork the salt-states repo and help us develop a truly large and scalable system for configuration management!

Notable Bug Fixes¶

Python 2.6 String Formatting¶

Python 2.6 does not support format strings without an index identifier, all of them have been repaired.

Cython Loading Disabled by Default¶

Cython loading requires a development tool chain to be installed on the minion, requiring this by default can cause problems for most Salt deployments. If Cython auto loading is desired it will need to be turned on in the minion config.

Salt 0.9.3 Release Notes¶

Salt 0.9.3 is finally arrived. This is another big step forward for Salt, new features range from proper FreeBSD support to fixing issues seen when attaching a minion to a master over the Internet.

The biggest improvements in 0.9.3 though can be found in the state system, it has progressed from something ready for early testers to a system ready to compete with platforms such as Puppet and Chef. The backbone of the state system has been greatly refined and many new features are available.

Download!¶

The Salt source can be downloaded from the salt GitHub site:

https://github.com/downloads/saltstack/salt/salt-0.9.3.tar.gz

Or from PyPI:

http://pypi.python.org/packages/source/s/salt/salt-0.9.3.tar.gz

For instructions on how to set up Salt please see the Installation instructions.

New Features¶

WAN Support¶

Recently more people have been testing Salt minions connecting to Salt Masters over the Internet. It was found that Minions would commonly loose their connection to the master when working over the internet. The minions can now detect if the connection has been lost and reconnect to the master, making WAN connections much more reliable.

State System Fixes¶

Substantial testing has gone into the state system and it is ready for real world usage. A great deal has been added to the documentation for states and the modules and functions available to states have been cleanly documented.

A number of State System bugs have also been founds and repaired, the output from the state system has also been refined to be extremely clear and concise.

Error reporting has also been introduced, issues found in sls files will now be clearly reported when executing Salt States.

Extend Declaration¶

The Salt States have also gained the extend declaration. This declaration allows for states to be cleanly modified in a post environment. Simply said, if there is an apache.sls file that declares the apache service, then another sls can include apache and then extend it:

include:
  - apache

extend:
  apache:
    service:
      - require:
        - pkg: mod_python

mod_python:
  pkg:
    - installed

The notable behavior with the extend functionality is that it literally extends or overwrites a declaration set up in another sls module. This means that Salt will behave as though the modifications were made directly to the apache sls. This ensures that the apache service in this example is directly tied to all requirements.

Highstate Structure Specification¶

This release comes with a clear specification of the Highstate data structure that is used to declare Salt States. This specification explains everything that can be declared in the Salt SLS modules.

The specification is extremely simple, and illustrates how Salt has been able to fulfill the requirements of a central configuration manager within a simple and easy to understand format and specification.

SheBang Renderer Switch¶

It came to our attention that having many renderers means that there may be a situation where more than one State Renderer should be available within a single State Tree.

The method chosen to accomplish this was something already familiar to developers and systems administrators, a SheBang. The Python State Renderer displays this new capability.

Python State Renderer¶

Until now Salt States could only be declared in yaml or json using Jinja or Mako. A new, very powerful, renderer has been added, making it possible to write Salt States in pure Python:

#!py

def run():
    '''
    Install the python-mako package
    '''
    return {'include': ['python'],
            'python-mako': {'pkg': ['installed']}}

This renderer is used by making a run function that returns the Highstate data structure. Any capabilities of Python can be used in pure Python sls modules.

This example of a pure Python sls module is the same as this example in yaml:

include:
  - python

python-mako:
  pkg:
    - installed

FreeBSD Support¶

Additional support has been added for FreeBSD, this is Salt's first branch out of the Linux world and proves the viability of Salt on non-Linux platforms.

Salt remote execution already worked on FreeBSD, and should work without issue on any Unix-like platform. But this support comes in the form of package management and user support, so Salt States also work on FreeBSD now.

The new freebsdpkg module provides package management support for FreeBSD and the new pw_user and pw_group provide user and group management.

Module and State Additions¶

Cron Support¶

Support for managing the system crontab has been added, declaring a cron state can be done easily:

date > /tmp/datestamp:
  cron:
    - present
    - user: fred
    - minute: 5
    - hour: 3

File State Additions¶

The file state has been given a number of new features, primarily the directory, recurse, symlink and absent functions.

file.directory

Make sure that a directory exists and has the right permissions.

/srv/foo:
  file:
    - directory
    - user: root
    - group: root
    - mode: 1755

file.symlink

Make a symlink.

/var/lib/www:
  file:
    - symlink
    - target: /srv/www
    - force: True

file.recurse

The recurse state function will recursively download a directory on the master file server and place it on the minion. Any change in the files on the master will be pushed to the minion. The recurse function is very powerful and has been tested by pushing out the full Linux kernel source.

/opt/code:
  file:
    - recurse
    - source: salt://linux

file.absent

Make sure that the file is not on the system, recursively deletes directories, files and symlinks.

/etc/httpd/conf.d/somebogusfile.conf:
  file:
    - absent

Sysctl Module and State¶

The sysctl module and state allows for sysctl components in the kernel to be managed easily. the sysctl module contains the following functions:

sysctl.show: Return a list of sysctl parameters for this minion
sysctl.get: Return a single sysctl parameter for this minion
sysctl.assign: Assign a single sysctl parameter for this minion
sysctl.persist: Assign and persist a simple sysctl parameter for this minion

The sysctl state allows for sysctl parameters to be assigned:

vm.swappiness:
  sysctl:
    - present
    - value: 20

Kernel Module Management¶

A module for managing Linux kernel modules has been added. The new functions are as follows:

kmod.available: Return a list of all available kernel modules
kmod.check_available: Check to see if the specified kernel module is available
kmod.lsmod: Return a dict containing information about currently loaded modules
kmod.load: Load the specified kernel module
kmod.remove: Unload the specified kernel module

The kmod state can enforce modules be either present or absent:

kvm_intel:
  kmod:
    - present

Ssh Authorized Keys¶

The ssh_auth state can distribute ssh authorized keys out to minions. Ssh authorized keys can be present or absent.

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:
  ssh_auth:
    - present
    - user: frank
    - enc: dsa
    - comment: 'Frank's key'

Salt 0.9.4 Release Notes¶

Salt 0.9.4 has arrived. This is a critical update that repairs a number of key bugs found in 0.9.3. But this update is not without feature additions as well! 0.9.4 adds support for Gentoo portage to the pkg module and state system. Also there are 2 major new state additions, the failhard option and the ability to set up finite state ordering with the order option.

This release also sees our largest increase in community contributions. These contributors have and continue to be the life blood of the Salt project, and the team continues to grow. I want to put out a big thanks to our new and existing contributors.

Download!¶

The Salt source can be downloaded from the salt GitHub site:

https://github.com/downloads/saltstack/salt/salt-0.9.4.tar.gz

Or from PyPI:

http://pypi.python.org/packages/source/s/salt/salt-0.9.4.tar.gz

For instructions on how to set up Salt please see the Installation instructions.

New Features¶

Failhard State Option¶

Normally, when a state fails Salt continues to execute the remainder of the defined states and will only refuse to execute states that require the failed state.

But the situation may exist, where you would want all state execution to stop if a single state execution fails. The capability to do this is called failing hard.

State Level Failhard¶

A single state can have a failhard set, this means that if this individual state fails that all state execution will immediately stop. This is a great thing to do if there is a state that sets up a critical config file and setting a require for each state that reads the config would be cumbersome. A good example of this would be setting up a package manager early on:

/etc/yum.repos.d/company.repo:
  file:
    - managed
    - source: salt://company/yumrepo.conf
    - user: root
    - group: root
    - mode: 644
    - order: 1
    - failhard: True

In this situation, the yum repo is going to be configured before other states, and if it fails to lay down the config file, than no other states will be executed.

Global Failhard¶

It may be desired to have failhard be applied to every state that is executed, if this is the case, then failhard can be set in the master configuration file. Setting failhard in the master configuration file will result in failing hard when any minion gathering states from the master have a state fail.

This is NOT the default behavior, normally Salt will only fail states that require a failed state.

Using the global failhard is generally not recommended, since it can result in states not being executed or even checked. It can also be confusing to see states failhard if an admin is not actively aware that the failhard has been set.

To use the global failhard set failhard: True in the master configuration

Finite Ordering of State Execution¶

When creating salt sls files, it is often important to ensure that they run in a specific order. While states will always execute in the same order, that order is not necessarily defined the way you want it.

A few tools exist in Salt to set up the correct state ordering, these tools consist of requisite declarations and order options.

The Order Option¶

Before using the order option, remember that the majority of state ordering should be done with requisite statements, and that a requisite statement will override an order option.

The order option is used by adding an order number to a state declaration with the option order:

vim:
  pkg:
    - installed
    - order: 1

By adding the order option to 1 this ensures that the vim package will be installed in tandem with any other state declaration set to the order 1.

Any state declared without an order option will be executed after all states with order options are executed.

But this construct can only handle ordering states from the beginning. Sometimes you may want to send a state to the end of the line, to do this set the order to last:

vim:
  pkg:
    - installed
    - order: last

Substantial testing has gone into the state system and it is ready for real world usage. A great deal has been added to the documentation for states and the modules and functions available to states have been cleanly documented.

A number of State System bugs have also been founds and repaired, the output from the state system has also been refined to be extremely clear and concise.

Error reporting has also been introduced, issues found in sls files will now be clearly reported when executing Salt States.

Gentoo Support¶

Additional experimental support has been added for Gentoo. This is found in the contribution from Doug Renn, aka nestegg.

Salt 0.9.5 Release Notes¶

Salt 0.9.5 is one of the largest steps forward in the development of Salt.

0.9.5 comes with many milestones, this release has seen the community of developers grow out to an international team of 46 code contributors and has many feature additions, feature enhancements, bug fixes and speed improvements.

Warning

Be sure to read the upgrade instructions about the switch to msgpack before upgrading!

Community¶

Nothing has proven to have more value to the development of Salt that the outstanding community that has been growing at such a great pace around Salt. This has proven not only that Salt has great value, but also the expandability of Salt is as exponential as I originally intended.

0.9.5 has received over 600 additional commits since 0.9.4 with a swath of new committers. The following individuals have contributed to the development of 0.9.5:

Aaron Bull Schaefer
Antti Kaihola
Bas Tichelaar
Brad Barden
Brian Wagner
Byron Clark
Chris Scheller
Christer Edwards
Clint Savage
Corey Quinn
David Boucha
Eivind Uggedal
Eric Poelke
Evan Borgstrom
Jed Glazner
Jeff Schroeder
Jeffrey C. Ollie
Jonas Buckner
Kent Tenney
Martin Schnabel
Maxim Burgerhout
Mitch Anderson
Nathaniel Whiteinge
Seth House
Thomas S Hatch
Thomas Schreiber
Tor Hveem
lzyeval
syphernl

This makes 21 new developers since 0.9.4 was released!

To keep up with the growing community follow Salt on Ohloh (http://www.ohloh.net/p/salt), to join the Salt development community, fork Salt on Github, and get coding (https://github.com/saltstack/salt)!

Major Features¶

SPEED! Pickle to msgpack¶

For a few months now we have been talking about moving away from Python pickles for network serialization, but a preferred serialization format had not yet been found. After an extensive performance testing period involving everything from JSON to protocol buffers, a clear winner emerged. Message Pack (http://msgpack.org/) proved to not only be the fastest and most compact, but also the most "salt like". Message Pack is simple, and the code involved is very small. The msgpack library for Python has been added directly to Salt.

This move introduces a few changes to Salt. First off, Salt is no longer a "noarch" package, since the msgpack lib is written in C. Salt 0.9.5 will also have compatibility issues with 0.9.4 with the default configuration.

We have gone through great lengths to avoid backwards compatibility issues with Salt, but changing the serialization medium was going to create issues regardless. Salt 0.9.5 is somewhat backwards compatible with earlier minions. A 0.9.5 master can command older minions, but only if the serial config value in the master is set to pickle. This will tell the master to publish messages in pickle format and will allow the master to receive messages in both msgpack and pickle formats.

Therefore the suggested methods for upgrading are either to just upgrade everything at once, or:

Upgrade the master to 0.9.5
Set serial to pickle in the master config
Upgrade the minions
Remove the serial option from the master config

Since pickles can be used as a security exploit the ability for a master to accept pickles from minions at all will be removed in a future release.

C Bindings for YAML¶

All of the YAML rendering is now done with the YAML C bindings. This speeds up all of the sls files when running states.

Experimental Windows Support¶

David Boucha has worked tirelessly to bring initial support to Salt for Microsoft Windows operating systems. Right now the Salt Minion can run as a native Windows service and accept commands.

In the weeks and months to come Windows will receive the full treatment and will have support for Salt States and more robust support for managing Windows systems. This is a big step forward for Salt to move entirely outside of the Unix world, and proves Salt is a viable cross platform solution. Big Thanks to Dave for his contribution here!

Dynamic Module Distribution¶

Many Salt users have expressed the desire to have Salt distribute in-house modules, states, renderers, returners, and grains. This support has been added in a number of ways:

Modules via States¶

Now when salt modules are deployed to a minion via the state system as a file, then the modules will be automatically loaded into the active running minion - no restart required - and into the active running state. So custom state modules can be deployed and used in the same state run.

Modules via Module Environment Directories¶

Under the file_roots each environment can now have directories that are used to deploy large groups of modules. These directories sync modules at the beginning of a state run on the minion, or can be manually synced via the Salt module salt.modules.saltutil.sync_all.

The directories are named:

_modules
_states
_grains
_renderers
_returners

The modules are pushed to their respective scopes on the minions.

Module Reloading¶

Modules can now be reloaded without restarting the minion, this is done by calling the salt.modules.sys.reload_modules function.

But wait, there's more! Now when a salt module of any type is added via states the modules will be automatically reloaded, allowing for modules to be laid down with states and then immediately used.

Finally, all modules are reloaded when modules are dynamically distributed from the salt master.

Enable / Disable Added to Service¶

A great deal of demand has existed for adding the capability to set services to be started at boot in the service module. This feature also comes with an overhaul of the service modules and initial systemd support.

This means that the service state can now accept - enable: True to make sure a service is enabled at boot, and - enable: False to make sure it is disabled.

Compound Target¶

A new target type has been added to the lineup, the compound target. In previous versions the desired minions could only be targeted via a single specific target type, but now many target specifications can be declared.

These targets can also be separated by and/or operators, so certain properties can be used to omit a node:

salt -C 'webserv* and G@os:Debian or E@db.*' test.ping

will match all minions with ids starting with webserv via a glob and minions matching the os:Debian grain. Or minions that match the db.* regular expression.

Node Groups¶

Often the convenience of having a predefined group of minions to execute targets on is desired. This can be accomplished with the new nodegroups feature. Nodegroups allow for predefined compound targets to be declared in the master configuration file:

nodegroups:
  group1: 'L@foo.domain.com,bar.domain.com,baz.domain.com and bl*.domain.com'
  group2: 'G@os:Debian and foo.domain.com'

And then used via the -N option:

salt -N group1 test.ping

Minion Side Data Store¶

The data module introduces the initial approach into storing persistent data on the minions, specific to the minions. This allows for data to be stored on minions that can be accessed from the master or from the minion.

The Minion datastore is young, and will eventually provide an interface similar to a more mature key/value pair server.

Major Grains Improvement¶

The Salt grains have been overhauled to include a massive amount of extra data. this includes hardware data, os data and salt specific data.

Salt -Q is Useful Now¶

In the past the salt query system, which would display the data from recent executions would be displayed in pure Python, and it was unreadable.

0.9.5 has added the outputter system to the -Q option, thus enabling the salt query system to return readable output.

Packaging Updates¶

Huge strides have been made in packaging Salt for distributions. These additions are thanks to our wonderful community where the work to set up packages has proceeded tirelessly.

FreeBSD¶

Salt on FreeBSD? There a port for that:

http://www.freebsd.org/cgi/cvsweb.cgi/ports/sysutils/salt/pkg-descr

This port was developed and added by Christer Edwards. This also marks the first time Salt has been included in an upstream packaging system!

Fedora and Red Hat Enterprise¶

Salt packages have been prepared for inclusion in the Fedora Project and in EPEL for Red Hat Enterprise 5 and 6. These packages are the result of the efforts made by Clint Savage (herlo).

Debian/Ubuntu¶

A team of many contributors have assisted in developing packages for Debian and Ubuntu. Salt is still actively seeking inclusion in upstream Debian and Ubuntu and the package data that has been prepared is being pushed through the needed channels for inclusion.

These packages have been prepared with the help of:

Corey
Aaron Toponce
and`

More to Come¶

We are actively seeking inclusion in more distributions. Primarily getting Salt into Gentoo, SUSE, OpenBSD and preparing Solaris support are all turning into higher priorities.

Salt 0.9.6 Release Notes¶

Salt 0.9.6 is a release targeting a few bugs and changes. This is primarily targeting an issue found in the names declaration in the state system. But a few other bugs were also repaired, like missing support for grains in extmods.

Due to a conflict in distribution packaging msgpack will no longer be bundled with Salt, and is required as a dependency.

New Features¶

HTTP and ftp support in files.managed¶

Now under the source option in the file.managed state a HTTP or ftp address can be used instead of a file located on the salt master.

Allow Multiple Returners¶

Now the returner interface can define multiple returners, and will also return data back to the master, making the process less ambiguous.

Minion Memory Improvements¶

A number of modules have been taken out of the minion if the underlying systems required by said modules are not present on the minion system. A number of other modules need to be stripped out in this same way which should continue to make the minion more efficient.

Minions Can Locally Cache Return Data¶

A new option, cache_jobs, has been added to the minion to allow for all of the historically run jobs to cache on the minion, allowing for looking up historic returns. By default cache_jobs is set to False.

Pure Python Template Support For file.managed¶

Templates in the file.managed state can now be defined in a Python script. This script needs to have a run function that returns the string that needs to be in the named file.

Salt 0.9.7 Release Notes¶

Salt 0.9.7 is here! The latest iteration of Salt brings more features and many fixes. This release is a great refinement over 0.9.6, adding many conveniences under the hood, as well as some features that make working with Salt much better.

A few highlights include the new Job system, refinements to the requisite system in states, the mod_init interface for states, external node classification, search path to managed files in the file state, and refinements and additions to dynamic module loading.

0.9.7 also introduces the long developed (and oft changed) unit test framework and the initial unit tests.

Major Features¶

Salt Jobs Interface¶

The new jobs interface makes the management of running executions much cleaner and more transparent. Building on the existing execution framework the jobs system allows clear introspection into the active running state of the running Salt interface.

The Jobs interface is centered in the new minion side proc system. The minions now store msgpack serialized files under /var/cache/salt/proc. These files keep track of the active state of processes on the minion.

Functions in the saltutil Module¶

A number of functions have been added to the saltutil module to manage and view the jobs:

running - Returns the data of all running jobs that are found in the proc directory.

find_job - Returns specific data about a certain job based on job id.

signal_job - Allows for a given jid to be sent a signal.

term_job - Sends a termination signal (SIGTERM, 15) to the process controlling the specified job.

kill_job Sends a kill signal (SIGKILL, 9) to the process controlling the specified job.

The jobs Runner¶

A convenience runner front end and reporting system has been added as well. The jobs runner contains functions to make viewing data easier and cleaner.

The jobs runner contains a number of functions...

active¶

The active function runs saltutil.running on all minions and formats the return data about all running jobs in a much more usable and compact format. The active function will also compare jobs that have returned and jobs that are still running, making it easier to see what systems have completed a job and what systems are still being waited on.

lookup_jid¶

When jobs are executed the return data is sent back to the master and cached. By default is is cached for 24 hours, but this can be configured via the keep_jobs option in the master configuration.

Using the lookup_jid runner will display the same return data that the initial job invocation with the salt command would display.

list_jobs¶

Before finding a historic job, it may be required to find the job id. list_jobs will parse the cached execution data and display all of the job data for jobs that have already, or partially returned.

External Node Classification¶

Salt can now use external node classifiers like Cobbler's cobbler-ext-nodes.

Salt uses specific data from the external node classifier. In particular the classes value denotes which sls modules to run, and the environment value sets to another environment.

An external node classification can be set in the master configuration file via the external_nodes option: http://salt.readthedocs.org/en/latest/ref/configuration/master.html#external-nodes

External nodes are loaded in addition to the top files. If it is intended to only use external nodes, do not deploy any top files.

State Mod Init System¶

An issue arose with the pkg state. Every time a package was run Salt would need to refresh the package database. This made systems with slower package metadata refresh speeds much slower to work with. To alleviate this issue the mod_init interface has been added to salt states.

The mod_init interface is a function that can be added to a state file. This function is called with the first state called. In the case of the pkg state, the mod_init function sets up a tag which makes the package database only refresh on the first attempt to install a package.

In a nutshell, the mod_init interface allows a state to run any command that only needs to be run once, or can be used to set up an environment for working with the state.

Source File Search Path¶

The file state continues to be refined, adding speed and capabilities. This release adds the ability to pass a list to the source option. This list is then iterated over until the source file is found, and the first found file is used.

The new syntax looks like this:

/etc/httpd/conf/httpd.conf:
  file:
    - managed
    - source:
      - salt://httpd/httpd.conf
      - http://myserver/httpd.conf: md5=8c1fe119e6f1fd96bc06614473509bf1

The source option can take sources in the list from the salt file server as well as an arbitrary web source. If using an arbitrary web source the checksum needs to be passed as well for file verification.

Initial Unit Testing Framework¶

Because of the module system, and the need to test real scenarios, the development of a viable unit testing system has been difficult, but unit testing has finally arrived. Only a small amount of unit testing coverage has been developed, much more coverage will be in place soon.

A huge thanks goes out to those who have helped with unit testing, and the contributions that have been made to get us where we are. Without these contributions unit tests would still be in the dark.

Compound Targets Expanded¶

Originally only support for and and or were available in the compound target. 0.9.7 adds the capability to negate compound targets with not.

Nodegroups in the Top File¶

Previously the nodegroups defined in the master configuration file could not be used to match nodes for states. The nodegroups support has been expanded and the nodegroups defined in the master configuration can now be used to match minions in the top file.

Salt 0.9.8 Release Notes¶

Salt 0.9.8 is a big step forward, with many additions and enhancements, as well as a number of precursors to advanced future developments.

This version of Salt adds much more power to the command line, making the old hard timeout issues a thing of the past and adds keyword argument support. These additions are also available in the salt client API, making the available API tools much more powerful.

The new pillar system allows for data to be stored on the master and assigned to minions in a granular way similar to the state system. It also allows flexibility for users who want to keep data out of their state tree similar to 'external lookup' functionality in other tools.

A new way to extend requisites was added, the "requisite in" statement. This makes adding requires or watch statements to external state decs much easier.

Additions to requisites making them much more powerful have been added as well as improved error checking for sls files in the state system. A new provider system has been added to allow for redirecting what modules run in the background for individual states.

Support for OpenSUSE has been added and support for Solaris has begun serious development. Windows support has been significantly enhanced as well.

The matcher and target systems have received a great deal of attention. The default behavior of grain matching has changed slightly to reflect the rest of salt and the compound matcher system has been refined.

A number of impressive features with keyword arguments have been added to both the CLI and to the state system. This makes states much more powerful and flexible while maintaining the simple configuration everyone loves.

The new batch size capability allows for executions to be rolled through a group of targeted minions a percentage or specific number at a time. This was added to prevent the "thundering herd" problem when targeting large numbers of minions for things like service restarts or file downloads.

Upgrade Considerations¶

Upgrade Issues¶

There was a previously missed oversight which could cause a newer minion to crash an older master. That oversight has been resolved so the version incompatibility issue will no longer occur. When upgrading to 0.9.8 make sure to upgrade the master first, followed by the minions.

Debian/Ubuntu Packages¶

The original Debian/Ubuntu packages were called salt and included all salt applications. New packages in the ppa are split by function. If an old salt package is installed then it should be manually removed and the new split packages need to be freshly installed.

On the master:

# apt-get purge salt
# apt-get install salt-{master,minion}

On the minions:

# apt-get purge salt
# apt-get install salt-minion

And on any Syndics:

# apt-get install salt-syndic

The official salt stack ppa for Ubuntu is located at: https://launchpad.net/~saltstack/+archive/salt

Major Features¶

Pillar¶

Pillar offers an interface to declare variable data on the master that is then assigned to the minions. The pillar data is made available to all modules, states, sls files etc. It is compiled on the master and is declared using the existing renderer system. This means that learning pillar should be fairly trivial to those already familiar with salt states.

CLI Additions¶

The salt command has received a serious overhaul and is more powerful than ever. Data is returned to the terminal as it is received, and the salt command will now wait for all running minions to return data before stopping. This makes adding very large --timeout arguments completely unnecessary and gets rid of long running operations returning empty {} when the timeout is exceeded.

When calling salt via sudo, the user originally running salt is saved to the log for auditing purposes. This makes it easy to see who ran what by just looking through the minion logs.

The salt-key command gained the -D and --delete-all arguments for removing all keys. Be careful with this one!

Running States Without a Master¶

The addition of running states without a salt-master has been added to 0.9.8. This feature allows for the unmodified salt state tree to be read locally from a minion. The result is that the UNMODIFIED state tree has just become portable, allowing minions to have a local copy of states or to manage states without a master entirely.

This is accomplished via the new file client interface in Salt that allows for the salt:// URI to be redirected to custom interfaces. This means that there are now two interfaces for the salt file server, calling the master or looking in a local, minion defined file_roots.

This new feature can be used by modifying the minion config to point to a local file_roots and setting the file_client option to local.

Keyword Arguments and States¶

State modules now accept the **kwargs argument. This results in all data in a sls file assigned to a state being made available to the state function.

This passes data in a transparent way back to the modules executing the logic. In particular, this allows adding arguments to the pkg.install module that enable more advanced and granular controls with respect to what the state is capable of.

An example of this along with the new debconf module for installing ldap client packages on Debian:

ldap-client-packages:
  pkg:
    - debconf: salt://debconf/ldap-client.ans
    - installed
    - names:
      - nslcd
      - libpam-ldapd
      - libnss-ldapd

Keyword Arguments and the CLI¶

In the past it was required that all arguments be passed in the proper order to the salt and salt-call commands. As of 0.9.8, keyword arguments can be passed in the form of kwarg=argument.

# salt -G 'type:dev' git.clone \
    repository=https://github.com/saltstack/salt.git cwd=/tmp/salt user=jeff

Requisite "in"¶

A new means to updating requisite statements has been added to make adding watchers and requires to external states easier. Before 0.9.8 the only way to extend the states that were watched by a state outside of the sls was to use an extend statement:

include:
  - http
extend:
  apache:
    service:
      - watch:
        - pkg: tomcat

tomcat:
  pkg:
    - installed

But the new Requisite in statement allows for easier extends for requisites:

include:
  - http

tomcat:
  pkg:
    - installed
    - watch_in:
      - service: apache

Requisite in is part of the extend system, so still remember to always include the sls that is being extended!

Providers¶

Salt predetermines what modules should be mapped to what uses based on the properties of a system. These determinations are generally made for modules that provide things like package and service management. The apt module maps to pkg on Debian and the yum module maps to pkg on Fedora for instance.

Sometimes in states, it may be necessary for a non-default module to be used for the desired functionality. For instance, an Arch Linux system may have been set up with systemd support. Instead of using the default service module detected for Arch Linux, the systemd module can be used:

http:
  service:
    - running
    - enable: True
    - provider: systemd

Default providers can also be defined in the minion config file:

providers:
  pkg: yumpkg5
  service: systemd

When default providers are passed in the minion config, then those providers will be applied to all functionality in Salt, this means that the functions called by the minion will use these modules, as well as states.

Requisite Glob Matching¶

Requisites can now be defined with glob expansion. This means that if there are many requisites, they can be defined on a single line.

To watch all files in a directory:

http:
  service:
    - running
    - enable: True
    - watch:
      - file: /etc/http/conf.d/*

This example will watch all defined files that match the glob /etc/http/conf.d/*

Batch Size¶

The new batch size option allows commands to be executed while maintaining that only so many hosts are executing the command at one time. This option can take a percentage or a finite number:

salt '*' -b 10 test.ping

salt -G 'os:RedHat' --batch-size 25% apache.signal restart

This will only run test.ping on 10 of the targeted minions at a time and then restart apache on 25% of the minions matching os:RedHat at a time and work through them all until the task is complete. This makes jobs like rolling web server restarts behind a load balancer or doing maintenance on BSD firewalls using carp much easier with salt.

Module Updates¶

This is a list of notable, but non-exhaustive updates with new and existing modules.

Windows support has seen a flurry of support this release cycle. We've gained all new file, network, and shadow modules. Please note that these are still a work in progress.

For our ruby users, new rvm and gem modules have been added along with the associated states

The virt module gained basic Xen support.

The yum pkg modules gained Scientific Linux support.

The pkg module on Debian, Ubuntu, and derivatives force apt to run in a non-interactive mode. This prevents issues when package installation waits for confirmation.

A pkg module for OpenSUSE's zypper was added.

The service module on Ubuntu natively supports upstart.

A new debconf module was contributed by our community for more advanced control over deb package deployments on Debian based distributions.

The mysql.user state and mysql module gained a password_hash argument.

The cmd module and state gained a shell keyword argument for specifying a shell other than /bin/sh on Linux / Unix systems.

New git and mercurial modules have been added for fans of distributed version control.

In Progress Development¶

Master Side State Compiling¶

While we feel strongly that the advantages gained with minion side state compiling are very critical, it does prevent certain features that may be desired. 0.9.8 has support for initial master side state compiling, but many more components still need to be developed, it is hoped that these can be finished for 0.9.9.

The goal is that states can be compiled on both the master and the minion allowing for compilation to be split between master and minion. Why will this be great? It will allow storing sensitive data on the master and sending it to some minions without all minions having access to it. This will be good for handling ssl certificates on front-end web servers for instance.

Solaris Support¶

Salt 0.9.8 sees the introduction of basic Solaris support. The daemon runs well, but grains and more of the modules need updating and testing.

Windows Support¶

Salt states on windows are now much more viable thanks to contributions from our community! States for file, service, local user, and local group management are more fully fleshed out along with network and disk modules. Windows users can also now manage registry entries using the new "reg" module.

Salt 0.9.9 Release Notes¶

0.9.9 is out and comes with some serious bug fixes and even more serious features. This release is the last major feature release before 1.0.0 and could be considered the 1.0.0 release candidate.

A few updates include more advanced kwargs support, the ability for salt states to more safely configure a running salt minion, better job directory management and the new state test interface.

Many new tests have been added as well, including the new minion swarm test that allows for easier testing of Salt working with large groups of minions. This means that if you have experienced stability issues with Salt before, particularly in larger deployments, that these bugs have been tested for, found, and killed.

Major Features¶

State Test Interface¶

Until 0.9.9 the only option when running states to see what was going to be changed was to print out the highstate with state.show_highstate and manually look it over. But now states can be run to discover what is going to be changed.

Passing the option test=True to many of the state functions will now cause the salt state system to only check for what is going to be changed and report on those changes.

salt '*' state.highstate test=True

Now states that would have made changes report them back in yellow.

State Syntax Update¶

A shorthand syntax has been added to sls files, and it will be the default syntax in documentation going forward. The old syntax is still fully supported and will not be deprecated, but it is recommended to move to the new syntax in the future. This change moves the state function up into the state name using a dot notation. This is in-line with how state functions are generally referred to as well:

The new way:

/etc/sudoers:
  file.present:
    - source: salt://sudo/sudoers
    - user: root
    - mode: 400

Use and Use_in Requisites¶

Two new requisite statements are available in 0.9.9. The use and use_in requisite and requisite-in allow for the transparent duplication of data between states. When a state "uses" another state it copies the other state's arguments as defaults. This was created in direct response to the new network state, and allows for many network interfaces to be configured in the same way easily. A simple example:

root_file:
  file.absent:
    - name: /tmp/nothing
    - user: root
    - mode: 644
    - group: root
    - use_in:
      - file: /etc/vimrc

fred_file:
  file.absent:
    - name: /tmp/nothing
    - user: fred
    - group: marketing
    - mode: 660

/files/marketing/district7.rst:
  file.present:
    - source: salt://marketing/district7.rst
    - template: jinja
    - use:
      - file: fred_file

/etc/vimrc:
  file.present:
    - source: salt://edit/vimrc

This makes the 2 lower state decs inherit the options from their respectively "used" state decs.

Network State¶

The new network state allows for the configuration of network devices via salt states and the ip salt module. This addition has been given to the project by Jeff Hutchins and Bret Palsson from Jive Communications.

Currently the only network configuration backend available is for Red Hat based systems, like Red Hat Enterprise, CentOS, and Fedora.

Exponential Jobs¶

Originally the jobs executed were stored on the master in the format: <cachedir>/jobs/jid/{minion ids} But this format restricted the number of jobs in the cache to the number of subdirectories allowed on the filesystem. Ext3 for instance limits subdirectories to 32000. To combat this the new format for 0.9.9 is: <cachedir>/jobs/jid_hash[:2]/jid_hash[2:]/{minion ids} So that now the number of maximum jobs that can be run before the cleanup cycle hits the job directory is substantially higher.

ssh_auth Additions¶

The original ssh_auth state was limited to accepting only arguments to apply to a public key, and the key itself. This was restrictive due to the way the we learned that many people were using the state, so the key section has been expanded to accept options and arguments to the key that over ride arguments passed in the state. This gives substantial power to using ssh_auth with names:

sshkeys:
  ssh_auth:
    - present
    - user: backup
    - enc: ssh-dss
    - options:
      - option1="value1"
      - option2="value2 flag2"
    - comment: backup
    - names:
      - AAAAB3NzaC1yc2EAAAABIwAAAQEAlyE26SMFFVY5YJvnL7AF5CRTPtAigSW1U887ASfBt6FDa7Qr1YdO5ochiLoz8aSiMKd5h4dhB6ymHbmntMPjQena29jQjXAK4AK0500rMShG1Y1HYEjTXjQxIy/SMjq2aycHI+abiVDn3sciQjsLsNW59t48Udivl2RjWG7Eo+LYiB17MKD5M40r5CP2K4B8nuL+r4oAZEHKOJUF3rzA20MZXHRQuki7vVeWcW7ie8JHNBcq8iObVSoruylXav4aKG02d/I4bz/l0UdGh18SpMB8zVnT3YF5nukQQ/ATspmhpU66s4ntMehULC+ljLvZL40ByNmF0TZc2sdSkA0111==
      - AAAAB3NzaC1yc2EAAAABIwAAAQEAlyE26SMFFVY5YJvnL7AF5CRTPtAigSW1U887ASfBt6FDa7Qr1YdO5ochiLoz8aSiMKd5h4dhB6ymHbmntMPjQena29jQjXAK4AK0500rMShG1Y1HYEjTXjQxIy/SMjq2aycHI+abiVDn3sciQjsLsNW59t48Udivl2RjWG7Eo+LYiB17MKD5M40r5CP2K4B8nuL+r4oAZEHKOJUF3rzA20MZXHRQuki7vVeWcW7ie8JHNBcq8iObVSoruylXav4aKG02d/I4bz/l0UdGh18SpMB8zVnT3YF5nukQQ/ATspmhpU66s4ntMehULC+ljLvZL40ByNmF0TZc2sdSkA0222== override
      - ssh-rsa AAAAB3NzaC1yc2EAAAABIwAAAQEAlyE26SMFFVY5YJvnL7AF5CRTPtAigSW1U887ASfBt6FDa7Qr1YdO5ochiLoz8aSiMKd5h4dhB6ymHbmntMPjQena29jQjXAK4AK0500rMShG1Y1HYEjTXjQxIy/SMjq2aycHI+abiVDn3sciQjsLsNW59t48Udivl2RjWG7Eo+LYiB17MKD5M40r5CP2K4B8nuL+r4oAZEHKOJUF3rzA20MZXHRQuki7vVeWcW7ie8JHNBcq8iObVSoruylXav4aKG02d/I4bz/l0UdGh18SpMB8zVnT3YF5nukQQ/ATspmhpU66s4ntMehULC+ljLvZL40ByNmF0TZc2sdSkA0333== override
      - ssh-rsa AAAAB3NzaC1yc2EAAAABIwAAAQEAlyE26SMFFVY5YJvnL7AF5CRTPtAigSW1U887ASfBt6FDa7Qr1YdO5ochiLoz8aSiMKd5h4dhB6ymHbmntMPjQena29jQjXAK4AK0500rMShG1Y1HYEjTXjQxIy/SMjq2aycHI+abiVDn3sciQjsLsNW59t48Udivl2RjWG7Eo+LYiB17MKD5M40r5CP2K4B8nuL+r4oAZEHKOJUF3rzA20MZXHRQuki7vVeWcW7ie8JHNBcq8iObVSoruylXav4aKG02d/I4bz/l0UdGh18SpMB8zVnT3YF5nukQQ/ATspmhpU66s4ntMehULC+ljLvZL40ByNmF0TZc2sdSkA0444==
      - option3="value3",option4="value4 flag4" ssh-rsa AAAAB3NzaC1yc2EAAAABIwAAAQEAlyE26SMFFVY5YJvnL7AF5CRTPtAigSW1U887ASfBt6FDa7Qr1YdO5ochiLoz8aSiMKd5h4dhB6ymHbmntMPjQena29jQjXAK4AK0500rMShG1Y1HYEjTXjQxIy/SMjq2aycHI+abiVDn3sciQjsLsNW59t48Udivl2RjWG7Eo+LYiB17MKD5M40r5CP2K4B8nuL+r4oAZEHKOJUF3rzA20MZXHRQuki7vVeWcW7ie8JHNBcq8iObVSoruylXav4aKG02d/I4bz/l0UdGh18SpMB8zVnT3YF5nukQQ/ATspmhpU66s4ntMehULC+ljLvZL40ByNmF0TZc2sdSkA0555== override
      - option3="value3" ssh-rsa AAAAB3NzaC1yc2EAAAABIwAAAQEAlyE26SMFFVY5YJvnL7AF5CRTPtAigSW1U887ASfBt6FDa7Qr1YdO5ochiLoz8aSiMKd5h4dhB6ymHbmntMPjQena29jQjXAK4AK0500rMShG1Y1HYEjTXjQxIy/SMjq2aycHI+abiVDn3sciQjsLsNW59t48Udivl2RjWG7Eo+LYiB17MKD5M40r5CP2K4B8nuL+r4oAZEHKOJUF3rzA20MZXHRQuki7vVeWcW7ie8JHNBcq8iObVSoruylXav4aKG02d/I4bz/l0UdGh18SpMB8zVnT3YF5nukQQ/ATspmhpU66s4ntMehULC+ljLvZL40ByNmF0TZc2sdSkA0666==

LocalClient Additions¶

To follow up the recent additions in 0.9.8 of additional kwargs support, 0.9.9 also adds the capability to send kwargs into commands via a dict. This addition to the LocalClient api can be used like so:

import salt.client

client = salt.client.LocalClient('/etc/salt/master')
ret = client.cmd('*', 'cmd.run', ['ls -l'], kwarg={'cwd': '/etc'})

This update has been added to all cmd methods in the LocalClient class.

Better Self Salting¶

One problem faced with running Salt states, is that it has been difficult to manage the Salt minion via states, this is due to the fact that if the minion is called to restart while a state run is happening then the state run would be killed. 0.9.9 slightly changes the process scope of the state runs, so now when salt is executing states it can safely restart the salt-minion daemon.

In addition to daemonizing the state run, the apt module also daemonizes. This update makes it possible to cleanly update the salt-minion package on Debian/Ubuntu systems without leaving apt in an inconsistent state or killing the active minion process mid-execution.

Wildcards for SLS Modules¶

Now, when including sls modules in include statements or in the top file, shell globs can be used. This can greatly simplify listing matched sls modules in the top file and include statements:

base:
  '*':
    - files*
    - core*

include:
  - users.dev.*
  - apache.ser*

External Pillar¶

Since the pillar data is just, data, it does not need to come expressly from the pillar interface. The external pillar system allows for hooks to be added making it possible to extract pillar data from any arbitrary external interface. The external pillar interface is configured via the ext_pillar option. Currently interfaces exist to gather external pillar data via hiera or via a shell command that sends yaml data to the terminal:

ext_pillar:
  - cmd_yaml: cat /etc/salt/ext.yaml
  - hiera: /etc/hirea.yaml

The initial external pillar interfaces and extra interfaces can be added to the file salt/pillar.py, it is planned to add more external pillar interfaces. If the need arises a new module loader interface will be created in the future to manage external pillar interfaces.

Single State Executions¶

The new state.single function allows for single states to be cleanly executed. This is a great tool for setting up a small group of states on a system or for testing out the behavior of single states:

salt '*' state.single user.present name=wade uid=2000

The test interface functions here as well, so changes can also be tested against as:

salt '*' state.single user.present name=wade uid=2000 test=True

New Tests¶

A few exciting new test interfaces have been added, the minion swarm allows not only testing of larger loads, but also allows users to see how Salt behaves with large groups of minions without having to create a large deployment.

Minion Swarm¶

The minion swarm test system allows for large groups of minions to be tested against easily without requiring large numbers of servers or virtual machines. The minion swarm creates as many minions as a system can handle and roots them in the /tmp directory and connects them to a master.

The benefit here is that we were able to replicate issues that happen only when there are large numbers of minions. A number of elusive bugs which were causing stability issues in masters and minions have since been hunted down. Bugs that used to take careful watch by users over several days can now be reliably replicated in minutes, and fixed in minutes.

Using the swarm is easy, make sure a master is up for the swarm to connect to, and then use the minionswarm.py script in the tests directory to spin up as many minions as you want. Remember, this is a fork bomb, don't spin up more than your hardware can handle!

python minionswarm.py -m 20 --master salt-master

Shell Tests¶

The new Shell testing system allows us to test the behavior of commands executed from a high level. This allows for the high level testing of salt runners and commands like salt-key.

Client Tests¶

Tests have been added to test the aspects of the client APIs and ensure that the client calls work, and that they manage passed data, in a desirable way.

Full Table of Contents¶

Frequently Asked Questions¶

Is Salt open-core?¶

What ports should I open on my firewall?¶

My script runs every time I run a state.highstate. Why?¶

When I run test.ping, why don't the Minions that aren't responding return anything? Returning False would be helpful.¶

How does Salt determine the Minion's id?¶

I'm using gitfs and my custom modules/states/etc are not syncing. Why?¶

Why aren't my custom modules/states/etc. available on my Minions?¶

Module X isn't available, even though the shell command it uses is installed. Why?¶

Introduction to Salt¶

The 30 second summary¶

Simplicity¶

Parallel execution¶

Building on proven technology¶

Python client interface¶

Fast, flexible, scalable¶

Open¶

Installation¶

Quick Install¶

Platform-specific installation instructions¶

Arch Linux¶

Installation¶

Stable Release¶

Tracking develop¶

Post-installation tasks¶

Debian Installation¶

Configure Apt¶

Squeeze (Old Stable)¶

Wheezy (Stable)¶

Sid (Unstable)¶

Import the repository key.¶

Update the package database¶

Install packages¶

Post-installation tasks¶

Notes¶

Fedora¶

Installation¶

Stable Release¶

Post-installation tasks¶

FreeBSD¶

Installation¶

Post-installation tasks¶

Gentoo¶

Post-installation tasks¶

OS X¶

Dependency Installation¶

Salt-Master Customizations¶

Post-installation tasks¶

RHEL / CentOS / Scientific Linux / Amazon Linux / Oracle Linux¶

Installation¶

Enabling EPEL on RHEL¶

Stable Release¶

Post-installation tasks¶

Solaris¶

Installation¶

Minion Configuration¶

Troubleshooting¶

Ubuntu Installation¶

Add repository¶

Install packages¶

Post-installation tasks¶

Windows¶

Windows Installer¶

Silent Installer option¶

Installer Source¶

Installation from source¶

Install on Windows XP 32bit¶

Single command bootstrap script¶

Packages management under Windows 2003¶

SUSE Installation¶

Installation¶

Stable Release¶

Post-installation tasks openSUSE¶

Post-installation tasks SLES¶

Unstable Release¶

openSUSE¶

Suse Linux Enterprise¶

Dependencies¶

Optional Dependencies¶

My script runs every time I run a `state.highstate`. Why?¶

When I run `test.ping`, why don't the Minions that aren't responding return anything? Returning `False` would be helpful.¶

Module `X` isn't available, even though the shell command it uses is installed. Why?¶

Matching the `minion id`¶