Configuration templating for SaltStack using Hierarchical substitution and Jinja
Project description
Pepa
Configuration templating for SaltStack using Hierarchical substitution and Jinja.
Pepa is part of the SaltStack as of release 2014.7.
Quick testing
You can easily test Pepa from the Command Line.
Create a virtual env. and install the required modules.
virtualenv venv
cd venv
source bin/activate
pip install pepa
Clone and run Pepa.
git clone https://github.com/mickep76/pepa.git
cd pepa
pepa -c examples/master test.example.com -d
Test and validate templates.
pepa-test --config examples/master -d
Look at output.
pepa-test --config examples/master -d -s
Install Pepa
git clone https://github.com/mickep76/pepa.git
mkdir -p /srv/salt/ext/pillar
cp pillar/pepa.py /srv/salt/ext/pillar/pepa.py
Configuring Pepa
extension_modules: /srv/salt/ext
ext_pillar:
- pepa:
resource: host # Name of resource directory and sub-key in pillars
sequence: # Sequence used for hierarchical substitution
- hostname: # Name of key
name: input # Alias used for template directory
base_only: True # Only use templates from Base environment, i.e. no staging
- default:
- environment:
- location..region:
name: region
- location..country:
name: country
- location..datacenter:
name: datacenter
- roles:
- osfinger:
name: os
- hostname:
name: override
base_only: True
subkey: True # Create a sub-key in pillars, named after the resource in this case [host]
subkey_only: True # Only create a sub-key, and leave the top level untouched
pepa_roots: # Base directory for each environment
base: /srv/pepa/base # Path for base environment
dev: /srv/pepa/base # Associate dev with base
qa: /srv/pepa/qa
prod: /srv/pepa/prod
# Use a different delimiter for nested dictionaries, defaults to '..' since some keys may use '.' in the name
#pepa_delimiter: ..
# Supply Grains for Pepa, this should **ONLY** be used for testing or validation
#pepa_grains:
# environment: dev
# Supply Pillar for Pepa, this should **ONLY** be used for testing or validation
#pepa_pillars:
# saltversion: 0.17.4
# Enable debug for Pepa, and keep Salt on warning
#log_level: debug
#log_granular_levels:
# salt: warning
# salt.loaded.ext.pillar.pepa: debug
Pepa can also be used in Master-less SaltStack setup.
Command line
usage: pepa [-h] [-c CONFIG] [-d] [-g GRAINS] [-p PILLAR] [-n] [-v]
hostname
positional arguments:
hostname Hostname
optional arguments:
-h, --help show this help message and exit
-c CONFIG, --config CONFIG
Configuration file
-r RESOURCE, --resource RESOURCE
Resource, defaults to first resource
-d, --debug Print debug info
-g GRAINS, --grains GRAINS
Input Grains as YAML
-p PILLAR, --pillar PILLAR
Input Pillar as YAML
-n, --no-color No color output
-v, --validate Validate output
Templates
Templates is configuration for a host or software, that can use information from Grains or Pillars. These can then be used for hierarchically substitution.
Example File: host/input/test_example_com.yaml
location..region: emea
location..country: nl
location..datacenter: foobar
environment: dev
roles:
- salt.master
network..gateway: 10.0.0.254
network..interfaces..eth0..hwaddr: 00:20:26:a1:12:12
network..interfaces..eth0..dhcp: False
network..interfaces..eth0..ipv4: 10.0.0.3
network..interfaces..eth0..netmask: 255.255.255.0
network..interfaces..eth0..fqdn: {{ hostname }}
cobbler..profile: fedora-19-x86_64
As you see in this example you can use Jinja directly inside the template.
Example File: host/region/amer.yaml
network..dns..servers:
- 10.0.0.1
- 10.0.0.2
time..ntp..servers:
- ntp1.amer.example.com
- ntp2.amer.example.com
- ntp3.amer.example.com
time..timezone: America/Chihuahua
yum..mirror: yum.amer.example.com
Each template is named after the value of the key using lowercase and all extended characters are replaced with underscore.
Example:
osfinger: Fedora-19
Would become:
fedora_19.yaml
Nested dictionaries
In order to create nested dictionaries as output you can use double dot “..” as a delimiter. You can change this using “pepa_delimiter” we choose double dot since single dot is already used by key names in some modules, and using “:” requires quoting in the YAML.
Example:
network..dns..servers:
- 10.0.0.1
- 10.0.0.2
network..dns..options:
- timeout:2
- attempts:1
- ndots:1
network..dns..search:
- example.com
Would become:
network:
dns:
servers:
- 10.0.0.1
- 10.0.0.2
options:
- timeout:2
- attempts:1
- ndots:1
search:
- example.com
Operators
Operators can be used to merge/unset a list/hash or set the key as immutable, so it can’t be changed.
Operator |
Description |
---|---|
merge() |
Merge list or hash |
unset() |
Unset key |
immutable() |
Set the key as immutable, so it can’t be changed |
imerge() |
Set immutable and merge |
iunset() |
Set immutable and unset |
Example:
network..dns..search..merge():
- foobar.com
- dummy.nl
owner..immutable(): Operations
host..printers..unset():
Testing
Pepa also come’s with a test/validation tool for templates. This allows you to test for valid Jinja/YAML and validate key values.
Command Line
usage: pepa-test [-h] [-c CONFIG] [-r RESOURCE] [-d] [-s] [-t] [-n]
optional arguments:
-h, --help show this help message and exit
-c CONFIG, --config CONFIG
Configuration file
-r RESOURCE, --resource RESOURCE
Configuration file, defaults to first resource
-d, --debug Print debug info
-s, --show Show result of template
-t, --teamcity Output validation in TeamCity format
-n, --no-color No color output
Test
A test is a set of input values for a template, it’s generally a good idea to create a separate test for each outcome if you have Jinja if statements.
Example: host/default/tests/default-1.yaml
grains..osfinger: Fedora-20
location..region: emea
You can also use Jinja inside a test, for example if you wan’t to iterate through test values.
Schema
A schema is a set of validation rules for each key/value. Schemas use Cerberus module for validation.
Example: host/schemas/pkgrepo.yaml
{% set hostname = '^([a-z0-9]([a-z0-9-]{0,61}[a-z0-9])?\.)+[a-zA-Z]{2,6}$' %}
{% set url = '(http|https?://([-\w\.]+)+(:\d+)?(/([\w/_\.]*(\?\S+)?)?)?)' %}
pkgrepo..mirror:
type: string
regex: {{ hostname }}
pkgrepo..type:
type: string
allowed: yum
pkgrepo..osabbr:
type: string
regex: ^(fc|rhel)[0-9]+$
{% for repo in [ 'base', 'everything', 'updates' ] %}
pkgrepo..repos..{{ repo }}..name:
type: string
regex: ^[A-Za-z\ 0-9\-\_]+$
pkgrepo..repos..{{ repo }}..baseurl:
type: string
regex: {{ url }}
{% endfor %}
You can also use Jinja inside a schema, for example if you wan’t to iterate through a list of different keys.
You can create complicated datastructures underneth a key, but it’s advisable to split it in several keys using the delimiter for a nested data structures.
Bad
network:
interfaces:
eth0:
ipv4: 192.168.1.2
netmask: 255.255.255.0
Good
network..interfaces..eth0..ipv4: 192.168.1.2
network..interfaces..eth0..netmask: 255.255.255.0
The first example you can’t properly use substitution and defining the schema becomes more complicated.
Project details
Release history Release notifications | RSS feed
Download files
Download the file for your platform. If you're not sure which to choose, learn more about installing packages.