We have separate job groups for aarch64 - "Fedora AArch64" and "Fedora AArch64 Updates".
I'm not actually sure there's any particularly great reason to do this. It means you have to look in different places for results. If we just got rid of these groups, you'd see the x86_64 and aarch64 results together, like the cloud results here. That seems...kinda better?
The home page would show just two groups, not four. I could look into whether we can tweak it to show more than three builds from each group, in that case.
Asset space is defined per group, so we'd lose the separate settings between x86_64 and aarch64 for that, but - I'm not sure that's actually good? It's probably better to have fewer groups with larger asset size limits for each group anyway.
@pwhalen , you're the one who set things up this way initially (a commit from you created the "Fedora AArch64" group) - do you remember if there was any particular reason why? Were you just copying what we did previously for PowerPC? Can you see a reason not to just merge the groups?
The "Fedora PowerPC" group was the first of these, I guess, in 2017, by Guy from IBM. We don't care a lot about Power lately, though. It's been disabled since the data centre move this summer and nobody's complained.
Thinking a bit more, I can see how this used to make sense but maybe doesn't any more.
For update tests especially, what we need to be able to do is easily and immediately see critical failures. We do that by looking at the Fedora Updates group and looking for red. So having PowerPC in that group doesn't make sense as Power is not a blocking arch so we won't ever gate updates on the Power tests; failures on Power are objectively less important so it makes sense to separate it out to keep its failures from 'polluting' the 'important' group.
The same was true for aarch64 for a long time, because while aarch64 is a blocking arch, the tests were flaky and we did not gate on them. But now we do. So it makes sense to have aarch64 failures show up in the 'important' group...
@lruzicka can you see any issues with this?
Update: I've deployed this on staging and it seems to be going fine so far. If it's alright there for a couple more days I'll deploy to prod. Getting rid of the old groups might be a bit tricky, though. Might require some database tinkering and/or upstream submissions (so we can hide the groups from the front page without outright deleting all their jobs, or something like that).
Oh, on a quick check of the source, looks like openQA by default only shows builds from the last 14 days on the front page. So once all jobs in the aarch64 groups are more than 14 days old, they won't appear on the front page any more. That should be fine, we can just go with that.
In fact, screw it, I'm merging this - if we leave it lying around it makes it complicated to land other test coverage extension efforts. If it causes problems somehow we'll roll it back.
This is done now.
Metadata Update from @adamwill: - Issue close_status updated to: Fixed - Issue status updated to: Closed (was: Open)
This issue has been migrated to Fedora Forge: https://forge.fedoraproject.org/quality/os-autoinst-distri-fedora/issues/447
Please continue any further discussion there.