#11156 rubygem-kgio in weird zombie state
Closed: Fixed 10 months ago by vondruch. Opened 2 years ago by vondruch.

The package seems to be in some zombie state. There is dead.package, however it is not marked as retired by dist-git pagure. I would not notice but this is one of the packages, where the maintainer was recently removed, so some automation apparently perceive the package being active.


Interesting

$ koji list-history --package=rubygem-kgio

Tue Feb  8 18:49:24 2022 package owner gomix set for rubygem-kgio in f37 by humaton
Tue Feb  8 18:49:24 2022 package list entry created: rubygem-kgio in f37 by humaton [still active]
Fri Mar 25 15:00:08 2022 package owner gomix set for rubygem-kgio in f36-Beta by humaton [still active]
Fri Mar 25 15:00:08 2022 package list entry created: rubygem-kgio in f36-Beta by humaton [still active]
Tue Aug  9 19:23:03 2022 package owner gomix set for rubygem-kgio in f38 by humaton
Tue Aug  9 19:23:03 2022 package list entry created: rubygem-kgio in f38 by humaton [still active]
Fri Sep  9 20:39:31 2022 package owner gomix set for rubygem-kgio in f37-Beta by humaton [still active]
Fri Sep  9 20:39:31 2022 package list entry created: rubygem-kgio in f37-Beta by humaton [still active]
Mon Nov 28 09:16:30 2022 package owner changed for rubygem-kgio in f37 by bodhi
    owner.name: gomix -> orphan
Mon Nov 28 11:45:42 2022 package owner changed for rubygem-kgio in f36 by bodhi
    owner.name: gomix -> orphan
Mon Nov 28 14:13:09 2022 package owner changed for rubygem-kgio in f35 by bodhi
    owner.name: gomix -> orphan
Mon Nov 28 17:58:27 2022 package owner changed for rubygem-kgio in module-package-list by bodhi
    owner.name: gomix -> orphan
Tue Nov 29 08:32:54 2022 package owner changed for rubygem-kgio in f38 by bodhi
    owner.name: gomix -> orphan

Metadata Update from @humaton:
- Issue tagged with: medium-gain, medium-trouble, ops

2 years ago

I have different, but very similar cases:

rubygem-boxgrinder-build
rubygem-boxgrinder-core

from this weeks meeting:

<nirik> so, bodhi is doing that somehow
[17:26:49] <nirik> Wed Apr 19 10:30:34 2023 package owner changed for rubygem-boxgrinder-build in module-package-list by bodhi
[17:26:49] <nirik>     owner.name: orphan -> releng
[17:26:49] <nirik> Thu Apr 20 10:32:06 2023 package owner changed for rubygem-boxgrinder-build in module-package-list by bodhi
[17:26:49] <nirik>     owner.name: releng -> orphan
[17:27:28] <jednorozec> yeah and on some packages it just added releng
[17:28:34] <nirik> was that branching day?
[17:28:41] <nirik> no. hum
[17:29:46] <jednorozec> nope 
[17:30:15] <jednorozec> I checked the original reported package and some of the changes are at randnom dates
[17:30:28] <nirik> so... it has to be the koji sync listener. somehow
[17:30:28] <jednorozec> some are relevant to branching or mass rebuild
[17:32:56] <nirik> ah ha
[17:33:07] <nirik> it is koji sync listener
[17:33:19] <jednorozec> what part of it?
[17:33:43] <nirik> https://pagure.io/releng/issue/6663 is referred to in a comment
[17:34:06] <nirik> basically it's syncing rawhide/rpms package list with module-package-list for mbs
[17:34:22] <nirik> but it must somehow not be seeing some things as retired when they are.
[17:34:39] <jednorozec> at least we have a suspect
[17:34:44] <nirik> roles/bodhi2/backend/templates/owner-sync-pagure.j2 is the script
[17:34:52] <nirik> yeah, I think it's this, but I am not sure why...

This needs more investigation.

Metadata Update from @patrikp:
- Issue assigned to patrikp

10 months ago

The package looks retired to me. Can this be closed?

The package looks retired to me. Can this be closed?

Seems to be in order now. However, I'd still be interested what steps you have take to check, to be double sure.

I.e. how to verify that a package is retired as opposed to orphaned?

You said "The package looks retired to me." and I wonder what is base for this claim? What were the action taken to address this issue. And what were the steps taken to verify this.

Retired packages have a dead.package file in the respective branches and are blocked in Koji. So I checked for the files and then ran koji list-pkgs --show-blocked --package=rubygem-kgio and can see that the Koji tags show [BLOCKED] as expected.

Retired packages have a dead.package file in the respective branches and are blocked in Koji.

Yep, it had dead.package already when I reported the issue.

So I checked for the files and then ran koji list-pkgs --show-blocked --package=rubygem-kgio and can see that the Koji tags show [BLOCKED] as expected.

It have not appeared retired in Pagure, but it appears now, so the issue is probably resolved for rubygem-kgio, while I am still not clear how that happened.

And how about the rubygem-boxgrinder-build and rubygem-boxgrinder-core?

And how about the rubygem-boxgrinder-build and rubygem-boxgrinder-core?

They were already retired but I am not sure how that happened. Making the dead.package file should be the thing that triggers blocking it in Koji but neither of the two packages had the file. Possibly they were blocked manually?

Anyway, I cloned both of the repositories and ran fedpkg retire 'reason' in the respective rawhide branches and the files are there now.

https://src.fedoraproject.org/rpms/rubygem-boxgrinder-core/c/897ed98be6c9d06e6d6f2f2ecf8777fec44d3e38?branch=rawhide

https://src.fedoraproject.org/rpms/rubygem-boxgrinder-build/c/2775602fa089065314f5721a1f02a4bf9fdfd658?branch=rawhide

Thx a lot. Closing this now.

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

10 months ago

Log in to comment on this ticket.

Metadata
Boards 1
Ops Status: Backlog