releng

#12013 container sync: fix expected container filenames for kiwi
Closed: Fixed by adamwill. Opened by adamwill.

Closed: Fixed
Reopen Issue
  • Describe the issue
    The container sync script has hardcoded expectations of what container image files will be called. These are no longer correct after the Kiwi transition. At least the Toolbox one is wrong, I have not checked the other two specifically yet.

At minimum we need to fix the script's expected filenames, but @kevin and I kicked around some ideas for revamping the process entirely to be...better. It could involve a message consumer (maybe a toddler) which listens out for compose complete messages and uses the productmd metadata to find the images it wants to sync (like how openQA schedules compose tests). We could also have a gating process where we don't sync till after some tests pass (we do have some testing of the toolbox image at least, at present). That could possibly be done via Bodhi, which does have the ability to handle container images as 'updates' and run skopeo on them.

  • When do you need this? (YYYY/MM/DD)
    ASAP at least to get Toolbox images syncing to the registries again.

  • When is this no longer needed or useful? (YYYY/MM/DD)
    When we stop wanting to publish container images to registries.

  • If we cannot complete your request, what is the impact?
    Registries will not get updated container images (at least Toolbox).

Mostly filing this to remind myself to work on it, I intend to do something for this today or over the weekend or on Monday. For reference see the "no such file or directory errors" in the compose cron output.
Edited by adamwill

Writing shell script without a safety net is "fun", but I think https://pagure.io/releng/pull-request/12014 should fix this. Think? Hope. One of those.

I'll test it out a bit when I get back to a real internet connection, I should be able to just turn the actual skopeo commands into test echos and see if it's doing the right things.

Metadata Update from @jnsamyak:
- Issue tagged with: dev, medium-gain, medium-trouble

Thanks for working on this, @adamwill & @kevin !

This should be fixed now, right? I see everything has been merged.

yes, you can see it working in the registries. https://quay.io/repository/fedora/fedora-toolbox?tab=history , https://quay.io/repository/fedora/fedora?tab=history , https://quay.io/repository/fedora/fedora-minimal?tab=history . The fact that you see recent entries with the hashes actually changing means that it is working.

Metadata Update from @adamwill:
- Issue close_status updated to: Fixed
- Issue status updated to: Closed (was: Open)

I would still like to rewrite this to be better in some ways, but...we'll see if I get the roundtuits. If anyone else wants to do it, feel free.
Metadata
None

dev medium-gain medium-trouble

None
None
None
Static archive. pagure.io was sunset in 2026; this is a read-only snapshot.
Red Hat Logo

Fedora is sponsored by Red Hat. Learn more about the relationship between Red Hat and Fedora »

© Red Hat, Inc. and others.