This project is a way of creating lots of config files from a Jinja2 template using a python script and an Ansible playbook. The process is:
- Write template with variables in place of IP addresses (for example)
- Gather variable values for the devices in an Excel spreadsheet
- Turn this into YAML using a python script
- Run the Ansible playbook to read the variables and the template, and produce config files.
- Add more csv files and templates and repeat.
Currently it creates seven (abbreviated) Juniper Network config files (three "routers" and four "srx") but given a different template could create anything.
Before doing anything else, you need to create at least one Jinja2-formatted
device template. This is a just a configuration file from whatever you are
working on, but with all the variables replaced by {{ item.VARIABLENAME }}.
There is a couple of examples supplied in the
roles/confgen2/templates/ directory
that should work for you and serve as a basis for your own template. Both are
for Junos config.
Once the template(s) are defined, you need to gather the variables for each device, such as hostname, IP addresses and so on. You may need to ask for this information from a customer, so it is easiest to do this in an Excel spreadsheet.
NOTE: make sure all your variable names are in lowercase, and use only underscores as a separator. Using hyphens is not supported by Ansible.
- Open the file
roles/confgen2/data/variables.xlsx - Create headings in row 1 for each of your variables specified in the template
- Fill in device 1's variable values in line 2
- Fill in device 2's variable values in line 3 and carry on
- If you have multiple Jinja2 templates, repeat the process across multiple tabs. Each tab can have different headings to match the template.
- When complete, save each tab as a
-variables.csvfile.
The Ansible playbook needs the variables in YAML, so a Python script is included to convert the CSV file you just generated into that format.
Run it using csv2yaml.py as follows to create roles/confgen/vars/main.yml
required for the next stage.
python csv2yaml.py -i roles/confgen2/data/*.csv -o roles/confgen2/vars/main.yml
We are now about to generate config files:
- First, delete the supplied config files in the
roles/confgen2/configs/*directories - these will be re-created when you run the playbook. - Type 'ansible-playbook confgen2.yml' to run the playbook.
- See the files re-appear.
This playbook operates locally on the script host - it does not push a config
out to a device.
The confgen.yml playbook simply calls a role called confgen2 and runs it against all hosts in the hosts file.
The tasks for the confgen2 role are in roles/confgen2/tasks/main.yml. This
reads in the source template, and writes a hostname.cfg config file for each
line in under srx_configs in the variables file. If you add new CSV files and
Jinja2 templates it'll be necessary to edit this file to add new tasks.
The templates are Jinja2 formatted files in roles/confgen2/templates/:
router-template.j2srx-template.j2
The variables are in roles/confgen2/vars/main.yml
Hopefully the python script will be sufficient for your needs, run with the "-h" argument for help:
$ python csv2yaml.py -h
usage: csv2yaml.py [-h] [-o [outFile]] [-i [inFile [inFile ...]]] [-debug]
Takes one or more CSV files, each containing device parameters, and turns them
into YAML for use by Ansible/Jinja2 config generation process.
optional arguments:
-h, --help show this help message and exit
-o [outFile] YAML file to write to
-i [inFile [inFile ...]]
CSV file(s) to read from
-debug Enable Debug Mode
This software is provided with no warranty. Based on work by Andrew Mulheirn.
https://github.com/amulheirn/confgen2
This file will convert each csv file provided as input to a YAML dictionary,
with each dictionary getting its name from the originating csv filename.
Any hyphens in the csv filename are converted to underscores in the dictionary
name. The outFile and inFile variables default to stdin and stdout
respectively, however reading from stdin is problematic.