Infrastructure SIG meeting 2024-01-25

Present: @evgeni (chair), @ekohl (notes), @ehelms, @Odilhao

Set up a Monitoring System · Issue #471 · theforeman/foreman-infra · GitHub

@evgeni did some basic setup that could monitor some basics. He was trying to also look at package updates, but it’s better to start monitoring basics and then expand on it. @Odilhao offered to help.

Once the EL8 node on Netways is installed, the EL7 node can be dropped and the resources used to set up a monitoring server.

Investigate Azure's FOSS credits program · Issue #1671 · theforeman/foreman-infra · GitHub

@cintrix84 is using this for foreman_azure_rm. @Odilhao will document the steps to onboard other developers and then we can move this issue to done.

move EL Jenkins nodes to EL8 · Issue #2020 · theforeman/foreman-infra · GitHub

Eric started to update EL9, but ran into problems with RVM (see Install openssl 1.0 on EL9 Jenkins nodes through RVM by ehelms · Pull Request #2016 · theforeman/foreman-infra · GitHub). First step will be to migrate existing nodes to EL8.

Plan to decommission the EL7 nodes:

  • Spin up 2 new Jenkins nodes on conova with EL8
  • Decommission node01.netways & node01.conova EL7 nodes
  • Turn off AWS nodes. The capacity isn’t needed right now, especially after we move more to GHA

Currently using CentOS Stream 8, even though it’s going EOL in May.
We aim to solve the RVM issue before then.

Replace RVM on Jenkins nodes · Issue #2021 · theforeman/foreman-infra · GitHub

This came from the RVM problems on EL9. Some options have been discussed (containers, plain bundler, rbenv) but no conclusion has been made.

Replace exim Puppet module (or exim) · Issue #1692 · theforeman/foreman-infra · GitHub

Module is now unused and could be dropped Drop exim module by ekohl · Pull Request #2005 · theforeman/foreman-infra · GitHub.

With CentOS Linux 7 going EOL in June, this is becoming more urgent.

Evgeni wants to split yum & deb into separate machines. Question is where to host the website and other parts hoest on it? Building used to have issues in GHA where it timed out. RSS is used to build stats. Perhaps a different solution for that.

@evgeni will do a write up