Investigate elasticsearch configuration using ansible

[philbudne]

Keeping notes about what I learn here...
The issue about cluster configuration is #365

[philbudne]

The first thing I found is an archived github repo: https://github.com/elastic/ansible-elasticsearch?tab=readme-ov-file
which doesn't support ES 8.x: https://github.com/elastic/ansible-elasticsearch/blob/main/docs/8x-support.md#what-is-not-working

The second thing I found is stuff to manage elastic (snapshots, reindex) using ansible, so not applicable in THIS situation:
https://galaxy.ansible.com/ui/repo/published/community/elastic/docs/
Tho it's possible https://galaxy.ansible.com/ui/repo/published/community/elastic/content/module/elastic_cluster_settings/
might do what we want/need [note: I don't think so... looks like run-time variables]

The third thing: https://github.com/geerlingguy/ansible-role-elasticsearch
"An Ansible Role that installs Elasticsearch on RedHat/CentOS or Debian/Ubuntu."
Author Information: This role was created in 2014 by Jeff Geerling Ansible for DevOps

[philbudne]

At Ansible Galaxy:

https://galaxy.ansible.com/ui/standalone/roles/ricardobarbosams/elasticsearch/
https://github.com/ricardobarbosams/elasticsearch (updated 7 months ago)
"This is a role to install Elasticsearch."
Requirements: Ubuntu 22 or previous [we're on Ubuntu 24]

https://galaxy.ansible.com/ui/standalone/roles/shaneholloman/elasticsearch/
An Ansible Role that installs Elasticsearch up to version 7.x on RedHat/CentOS or Debian/Ubuntu.
This role was created in 2023

https://galaxy.ansible.com/ui/standalone/roles/claranet/elasticsearch/
https://github.com/claranet/ansible-role-elasticsearch (updated 8 months ago)
Saw has updates for Ubuntu 24, but nothing (yet) about ES v8 (default elasticsearch_version: 7.17.6)

[philbudne]

Looking at forks of archived elastic role (the first link in this issue):
https://github.com/elastic/ansible-elasticsearch/forks
there are 857 forks(!!), 16 modified in the last year:
https://github.com/elastic/ansible-elasticsearch/forks?include=active&page=1&period=1y&sort_by=stargazer_counts

Searching for elasticsearch 8 ansible:

YouTube: ElasticSearch 8.x installation on Ubuntu 20.04 LTS with Ansible
Using https://github.com/0nurulusoy/es-ansible-installation (?)

https://medium.com/@abdullah.pelit/installation-of-elk-stack-version-8-9-with-ansible-aaec8a0f2765

[philbudne]

Of the forks of the original elastic ansible role (first link), these seem to have been updated for 8.x:
https://github.com/thuync/ansible-elasticsearch
https://github.com/ES-MAD-GEMMA-ROADMAP/ansible-elasticsearch

Unclear how many (if any) changes are ACTUALLY needed for v8!
(but as mentioned above, updates for ansible changes are needed)

Many have changes for changes to ansible "include" directive.

[pgulley]

Great work compiling resources! So it seems we're stuck using a fork of ansible to get this done- I say just pick one that looks good, I'm not sure what input I have that might help differentiate, other than it should probably be minimally different from ansible core so we can keep up with changes there.

[philbudne]

The repos I'm looking at aren't forks of ansible proper, but rather add-ins to ansible. ansible has an ecosystem called "ansible galaxy" of add-ins, but I haven't spotted the candidates (last two links above) that seem most promising....

Since we're looking at using someone's github fork (almost certainly of the "role" that elastic published, then dropped support for), I'm wondering if we'll want to make a sub-fork into the mediacloud project, so we're not at risk of having something we depend on disappear or break....

I've put together a "task" to install the agent to report system stats to grafana, testing against a Docker container, and have worked on collecting and arranging the parameters for configuring ES, getting a quick reminder of all the things I hate about ansible (primarily yaml and jinja2 related, but also that it expects data structures to be layed out JUST so), edging closer to trying to install ES into athe container..

[pgulley]

Nice work getting a draft up in #367- @thepsalmist, can you take a look at that PR when you get a chance? I think you have the most direct experience setting up ES clusters so I'd love for you to take the lead in making sure we've covered our bases with the main setup procedure here.

[pgulley]

Hey Xavier! Just wanted to ask for a status update here- do you think we'll have something working here before next wednesday?