There have been no updates posted on Discourse since the last meeting.
No progress - convert remaining tasks to multiple issues
CentOS 8 has been archived on January 31st which means the mirror network no longer has it. We’ve updated our infra to use Vault for now.
Patrick has built a server to store RHEL but due to time constraints no progress has been made on actually using RHEL within Koji.
No direct progress but there’s discussion if we continue using Duffy at all. If we don’t, the whole direction may shift. This has to with the Duffy infra being out of warranty and it has to move to another data center. CentOS CI is investigating if it can avoid doing so and retire the hardware instead. Alternatives are looked at. @evgeni has sent a summary of our needs.
Since the last meeting the new job did run and it mostly worked well. There were a few issues:
- Jenkins doesn’t recognize the author (GitHub Actions) and means we manually need to kick off CI. There is configuration to set the allowed users. Currently this is only set the Foreman organization.
- One false positive: ruby-foreman-deface. It’s not really a plugin but a special packaging so plugins can use the same dependency. It’s probably easiest to bump it rather than to special case it.
- There was also a failure with rubygem-foreman_puppet. This is due to an older version of python-semver in our CI, which is installed via Debian repositories. This was the only plugin that failed but that was because it was the only bumped plugin with a package.json, which is where python-semver is used. We either need to make the script compatible with an older python-semver or make sure we install the latest version.
Follow up idea: in stable branches only pick up to the latest .z version. However, it’s hard to say it’s really safe. Not all plugins follow semver. We generally also cherry pick but some plugins do have stable branches that are actually maintained. It may be possible to run a workflow with explicit parameters, like branch and plugin(s). This is still manual work, but it’s better to solve the problem partially than try to aim for everything.
To complete this item we’ll need to have it scheduled. All other items should be separate issues.
Today our Jenkins nodes run EL7 (CentOS 7 / RHEL 7). During the discussion it is suggested to only use CentOS 8 Stream in the future.
Agreed to split up the issue into a milestone and various more specific issues.
Assigned to @ekohl and moved to In Progress.
@pcreech has started to look into this so it has been assigned and moved to In Progress. It won’t make it in time for Foreman 3.2, but possibly 3.3.
Resolved and thus archived.