Skip to main content

Automation wrapper for bash and python commands

Project description


Automation wrapper for bash and python commands


automatix is a wrapper for scripted sysadmin tasks. It offers some useful functionality for easier scripting and having full control over the automated process.

The idea of automatix is to write down all the commands you would normally type to your commandline or python console into a YAML file. Then use automatix to execute these commands.

There are different modes for automatix to work. Without any parameters automatix will try to execute the specified command pipeline from the script file until an error occurs or the pipeline is done. The interactive mode (-i) asks for every single commandline step whether to execute, skip or abort. Forced mode (-f) will also proceed if errors occur.

automatix is originally designed for internal //SEIBERT/MEDIA use. It comes therefore with bundlewrap and teamvault support as well as the possibility to use your own logging library.


Beware that this tool cannot substitute the system administrators brain and it needs a responsible handling, since you can do (and destroy) almost everything with it.

Automatix evaluates YAML files and executes defined commands as shell or python commands. There is no check for harmful commands. Be aware that this can cause critical damage to your system.

Please use the interactive mode and doublecheck commands before executing. Usage of automatix is at your own risk!


Automatix requires Python ≥ 3.6.

pip install automatix


You can specify a path to a configuration YAML file via the environment variable AUTOMATIX_CONFIG. Default location is "~/.automatix.cfg.yaml".


# Path to scripts directory
script_dir: ~/automatix_script_files

# Global constants for use in pipeline scripts
  apt_update: 'apt-get -qy update'
  apt_upgrade: 'DEBIAN_FRONTEND=noninteractive apt-get -qy -o Dpkg::Options::=--force-confold --no-install-recommends upgrade'
  apt_full_upgrade: 'DEBIAN_FRONTEND=noninteractive apt-get -qy -o Dpkg::Options::=--force-confold --no-install-recommends full-upgrade'

# Encoding
encoding: utf-8

# Path for shell imports
import_path: '.'

# SSH Command used for remote connections
ssh_cmd: 'ssh {hostname} sudo '

# Temporary directory on remote machines for shell imports
remote_tmp_dir: 'automatix_tmp'

# Logger
logger: mylogger

# Logging library (has to implement the init_logger method)
logging_lib: mylib.logging

# Bundlewrap support, bundlewrap has to be installed (default: false)
bundlewrap: true

# Teamvault / Secret support, bundlewrap-teamvault has to be installed (default: false)
teamvault: true


automatix [--help|-h] [--systems [SYSTEM1=NODENAME ...]] [--vars [VAR1=VALUE1 ...]] [--secrets [SECRET1=SECRETID ...]] [--print-overview|-p] [--jump-to|-j JUMP_TO] [--interactive|-i] [--force|-f] [--debug|-d] [--] scriptfile


scriptfile : The only required parameter for this tool to work. Use " -- " if needed to delimit this from argument fields. See SCRIPTFILE section for more information.

-h, --help : View help message and exit.

--systems SYSTEM1=NODENAME : Use this to set systems without adding them to the config or to overwrite them. You can specify multiple systems like: --systems v1=string1 v2=string2 v3=string3

--vars VAR1=VALUE1 : Use this to set vars without adding them to the config or to overwrite them. You can specify multiple vars like: --vars v1=string1 v2=string2 v3=string3

--secrets SECRET1=SECRETID : Use this to set secrets without adding them to the config or to overwrite them. You can specify multiple secrets like: --secrets v1=string1 v2=string2 v3=string3 (only if teamvault is enabled)

--print-overview, -p : Just print command pipeline overview with indices then exit without executing the commandline. Note that the always pipeline will be executed anyway.

--jump-to JUMP_TO, -j JUMP_TO : Jump to step with index JUMP_TO instead of starting at the beginning. Use -p or the output messages to determine the desired step index. You can use negative numbers to start counting from the end.

--interactive, -i : Confirm actions before executing.

--force, -f : Try always to proceed (except manual steps), even if errors occur (no retries).

--debug, -d : Activate debug log level.


The scriptfile describes your automated process. Therefore it contains information about systems, variables, secrets and the command pipeline.

You can provide a path to your scriptfile or place your scriptfile in the predefined directory (see CONFIGURATION section). The path has precedence over the predefined directory, if the file exists at both locations.

The scriptfile has to contain valid YAML.

EXAMPLE: scriptfile

name: Migration Server XY
# Systems you like to refer to in pipeline (accessible via 'system_source')
# If Bundlewrap support is activated use node names instead of hostnames.
# Custom vars to use in pipeline
  version: 1.2.3
  domain: 'bla.mein-test-system'
# Teamvault Secrets, if activated (left: like vars, right: SECRETID_FIELD, FIELD=username|password|file)
  web_user: v6GQag_username
  web_pw: v6GQag_password
# Imports for functions you like to use (path may be modified in configuration)
# like command pipeline but will be exectuted always beforehand
  - python: global nc; import mylib as nc
  - remote@target: systemctl stop server
  - remote@source: zfs snapshot -r tank@before-migration
  - manual: Please trigger preparing tasks via webinterface
  - myvar=local: curl -L -vvv -k https://{domain}/
  - local: echo " {domain}" >> /etc/hosts
  - sla=python: source_node.metadata.get('sla')
  - python: |
        sla = '{sla}'
        if sla == 'gold':
            print('Wow that\'s pretty cool. You have SLA Gold.')
            print('Oh. Running out of money? SLA Gold is worth it. You should check your wallet.')
  - local: rm temp_files


name (string) : Just a name for the process. Does not do anything.

systems (associative array) : Define some systems. Value has to be an valid and existing bundlewrap nodename. You can refer to these systems in the command pipeline in multiple ways:

  1. remote@systemname as your command action (see below)

  2. via {system_systemname} which will be replaced with the value

  3. via systemname_node in python actions to use the bw node object

vars (associative array) : Define some vars. These are accessible in the command pipeline via {varname}.

secrets (associative array) : Define teamvault secrets. Value has to be in this format: SECRETID_FIELD. FIELD must be one of username, password or file. The resolved secret values are accessible in command line via {secretname}. (only if teamvault is enabled)

imports (list) : Listed shell files (see CONFIGURATION section) will be sourced before every local or remote command execution. For remote commands, these files are transferred via tar and ssh to your home directory on the remote system beforehand and deleted afterwards. This is meant to define some functions you may need.

always, cleanup (list of associative arrays) : See ALWAYS / CLEANUP PIPELINE section.

pipeline (list of associative arrays) : See PIPELINE section.


Here you define the commands automatix shall execute.

KEY: One of these possible command actions:

  1. manual: Some manual instruction for the user. The user has to confirm, that automatix may proceed.

  2. local: Local shell command to execute. Imports will be sourced beforehand. /bin/bash will be used for execution.

  3. remote@systemname: Remote shell command to execute. Systemname has to be a defined system. The command will be run via SSH (without pseudo-terminal allocation). It uses the standard SSH command. Therefore your .ssh/config should be respected.

  4. python: Python code to execute. If bundlewrap is enabled, system node objects are avaiable via systemname_node.

ASSIGNMENT: For local, remote and python action you can also define a variable to which the output will be assigned. To do this prefix the desired variablename and = before the action key, e.g. myvar=python: system_node.hostname. Be careful when working with multiline statements. In python the first line is likely to set the variable. All variables will be converted to strings when used to build commands in following steps.

VALUE: Your command. Variables will be replaced with Python format function. Therefore use curly brackets to refer to variables, systems, secrets and constants.

Constants are available via const_KEY, where KEY is the dictionary key in tools/automatix/ There you can define some widely used constants.

In most cases its a good idea to define your command in quotes to avoid parsing errors, but it is not always necessary. Another way is to use '|' to indicate a literal scalar block. There you can even define whole program structures for python (see example).

Escaping in Pipeline

Because automatix uses Python's format() function:
{ -> {{
} -> }}

Standard YAML escapes (see also
' -> ''
" -> \"
\ -> \\
: -> Please use quotes (double or single).


Same usage as the 'normal' command pipeline, but will be executed every time at start of automatix (always) or at the end (cleanup) even if aborted (a). The commands are executed without --interactive flag, independend of the specified parameters.

Intended use case for always: python imports or informations that are needed afterwards and do not change anything on systems. You want to have these available even if using --jump|-j feature.

Intended use case for cleanup: Remove temporary files or artifacts.


AUTOMATIX_CONFIG: Specify the path to the configuration file. Default is "~/.automatix.cfg.yaml".

ENCODING: Specify output encoding. Default is "UTF-8".

Additionally you can modify the environment to adjust things to your needs.


For python action you can import libraries globally, e.g. global pb; import isaclib.ionos as pb. This allows you to assign output of library functions to (automatix) variables. BUT CAUTION: Choosing already existing (Python) variable names may lead to unexpected behaviour!!! Maybe you want to check the source code (
(Explanation: automatix is written in Python and uses 'exec' to execute the command in function context. If you declare variables globally they remain across commands.)

For python action besides the Bundlewrap node objects (SYSTEMNAME _node) there are some modules, constants and functions which are already imported:
re, subprocess, quote(from shlex)


There are different ways to start scripting with automatix. The author's approach is mainly to consider the process and simply write down, what to do (manual steps for complex or not automated steps) and which commands to use.
Then start automatix in interactive mode (-i) and adjust the single steps one by one. Replace manual steps, if suitable. Whenever adjustment is needed, abort, adjust and restart automatix with jump (-j) to the adjusted step.
Repeat this procedure to automate more and more and increase quality, whenever you feel like it.

Consider to put often used paths or code sequences in automatix variables for better readability.
Do the same with variable content like URLs, to make it possible to overwrite it by command line options. Where ever possible prefer to use functions to determine already available information, such as BW metadata, instead of defining things explicitly. This will make things easier when using the script with different systems / parameters.

Preferred way of using automatix is to put often used and complex algorithms in shell functions or python libraries (shelllib/pylib) and import them. Advantage of this approach is that you can use your implemented functions multiple times and build up a toolbox of nice functionality over time.


Manual steps will always cause automatix to stop and wait for user input.

Be careful with assignments containing line breaks (echo, ...). Using the variables may lead to unexpected behaviour or errors.

Assignments containing null bytes are currently not supported.

Because the always pipeline should not change anything, aborting while running this pipeline will not trigger a cleanup.

If you want to abort the pipeline without triggering the cleanup pipeline, use CRTL+C.

While aborting remote functions (via imports), automatix is not able to determine still running processes invoked by the function, because it only checks the processes for the commands (in this case the function name) which is called in the pipeline.

Project details

Download files

Download the file for your platform. If you're not sure which to choose, learn more about installing packages.

Source Distribution

automatix-0.1.0.tar.gz (17.3 kB view hashes)

Uploaded Source

Built Distribution

automatix-0.1.0-py3-none-any.whl (15.3 kB view hashes)

Uploaded Python 3

Supported by

AWS AWS Cloud computing and Security Sponsor Datadog Datadog Monitoring Fastly Fastly CDN Google Google Download Analytics Microsoft Microsoft PSF Sponsor Pingdom Pingdom Monitoring Sentry Sentry Error logging StatusPage StatusPage Status page