Now that MBS is back online, I've done a bunch of flatpak builds and noticed the following:
Yesterday, the day after I had submitted https://bodhi.fedoraproject.org/updates/FEDORA-FLATPAK-2020-b63ed36598 to testing, https://bodhi.fedoraproject.org/composes/ showed a "Fedora 32 Flatpaks testing compose" push issue. I didn't write down the exact error, but it was something to do with bodhi-skopeo-lite failing to copy the image somewhere.
@mohanboddu must have resumed the push as the error was gone later yesterday. However, the updates did not appear when doing a 'flatpak update' from the testing remote, so something must have gone wrong in the process.
Today, I see a similar error about "Fedora 31 Flatpaks testing compose" when looking at https://bodhi.fedoraproject.org/composes/F31F/testing. This time the error message is:
/usr/bin/bodhi-skopeo-lite copy docker://candidate-registry.fedoraproject.org/gimp:master-3120200731114206.1 docker://registry.fedoraproject.org/gimp:master-3120200731114206.1 returned a non-0 exit code: 1
Can someone look at the logs and diagnose what's going wrong here, please? There must be an actual error message what's going wrong with the copy. I don't want to silently lose flatpak builds if something is going wrong with updating the registry.
(I'll note that I'll be mostly gone for two weeks, but @otaylor might be able with the flatpak side of things if needed.)
Note that this could be a dup of https://pagure.io/fedora-infrastructure/issue/9177 , this time just bodhi occasionally hitting it.
Yeah, it could be 9177... I manually ran a copy and it worked fine. I then resumed the failed push and... it worked fine. ;(
To be clear the f31-flatpak-updates-testing push failed yesterday, so I resumed it and it completed fine.
I think we are back on track aside from possibly hitting this again with 9177. ;(
Metadata Update from @kevin: - Issue close_status updated to: Fixed - Issue status updated to: Closed (was: Open)
OK, thanks. There is something going wrong with flatpaks as in they don't appear in the testing remote. I thought it might be related to the copy issue but it could be something unrelated (maybe the indexer is broken).
I'm on PTO and mostly offline for two weeks; I'll investigate what's going on when I'm back.
I fixed this issue, was 'too many open files' on oci-registry02.
Please let me know if you still see it...
Excellent! Thanks, @kevin!