#9759 fedpkg request-branch f31 fails: Invalid body, keys: sls missing
Closed: Fixed a month ago by humaton. Opened 11 months ago by alebastr.

  • Describe the issue
    Since yesterday fedpkg request-branch f31 tickets are failing with cryptic message 'Invalid body, keys: sls missing'.
    The error is unique enough to match something in my memory: https://lists.fedoraproject.org/archives/list/devel@lists.fedoraproject.org/thread/FWDHEPS6Y2S7UHLWSW5SVN5RYEDRFPNV/
    Based on the responses in the thread, I'd make a guess that yesterday's untimely EOL of f31 in pdc may have something to do with that. Of course, being totally clueless on Fedora infra I might just make a wrong assumption :)

  • When do you need this? (YYYY/MM/DD)
    Before the real f31 EOL, I guess.

  • When is this no longer needed or useful? (YYYY/MM/DD)

  • If we cannot complete your request, what is the impact?
    No new packages in F31.


Metadata Update from @humaton:
- Issue tagged with: dev, low-gain, low-trouble

11 months ago

But do we really want new packages getting into a release that is going to be EOL in a couple of weeks?

I will include a more meaning full message in the tool used to process tickets.

But do we really want new packages getting into a release that is going to be EOL in a couple of weeks?

Yes, we do. It is not EOL yet.

Ok, PR: https://pagure.io/fedscm-admin/pull-request/50

I will close this issue once f32 bodhi update with fedscm_admin is out.

@humaton, thanks for fixing that!

But do we really want new packages getting into a release that is going to be EOL in a couple of weeks?

fedpkg request-branch --all-releases takes a list of active branches from pdc and creates branch requests for all of them. If other tools have different opinion on which branches are active, it results a bad packager experience.
Speaking of pdc, is it intentional that fedscm-admin does not use the same API to check branch status and keeps internal list of eol dates instead? Single source of truth would decrease the chance of anyone else getting this message :smile:

I will include a more meaning full message in the tool used to process tickets.

That would be awesome. Especially if it could contain something understandable for a packager, possibly with suggestion for further actions.

I will close this issue once f32 bodhi update with fedscm_admin is out.

At which point I can start reopening f31 tickets? Someone with updated fedscm-admin would have to reprocess them and I guess this should happen after the update reaches stable.

note that by policy we already stop accepting f31 branches when f33 is released (ie, there's 1 month of support left for f31).

note that by policy we already stop accepting f31 branches when f33 is released (ie, there's 1 month of support left for f31).

Oh. That actually requires the list of dates in fedscm-admin. /product-versions pdc endpoint doesn't provide enough info to figure that out.
Looking at the fedpkg source, it doesn't seem to be aware of that policy though. Is it possible to tweak --all-releases to take that into account?

@alebastr What is the package you are trying to request for?

@mohanboddu releng/fedora-scm-requests#29108 and releng/fedora-scm-requests#29135.
But actually this fails for all f31 branch requests since 29108, currently 8 tickets.

Edit: and a few more previously reopened, such as releng/fedora-scm-requests#28909. https://pagure.io/releng/fedora-scm-requests/issues?status=Invalid&search_pattern=f31&order=desc&order_key=closed_at with anything closed after 2020-09-16.

note that by policy we already stop accepting f31 branches when f33 is released (ie, there's 1 month of support left for f31).

What policy is this?

That says:

Branches for new packages in the SCM are not allowed for distribution X after the Fedora X+2 release and new builds are no longer allowed.

That could be either interpreted as:

Branches for new packages in the SCM are not allowed for distribution X immediately after the Fedora X+2 release and new builds are no longer allowed.

Or:

Branches for new packages in the SCM are not allowed for distribution X some time after the Fedora X+2 release and new builds are no longer allowed.

I don't see how that would say:

Branches for new packages in the SCM are not allowed for distribution X immediately after the Fedora X+2 release and new builds are no longer allowed some time after.

(I am not a native speaker, but your interpretation is weird to me.)

End of Life is defined as Final Release (GA) of next-but-one release plus four weeks -- that AFAIK applies both to new packages and any builds/updates of the existing packages.

Well, perhaps thats worded poorly, but what we have been doing for the last... 10+ years? is:

  • at release of fedora N, new branches for N-2 is no longer allowed.
  • N-2 is EOL 4 weeks later.

This was the case in fedscm-admin, in pkgdb before it, and in the manual scripting before that.

If you want to change it, feel free to propose that as a change to FESCo.
If you want to improve docs, I'm all for it.

I don't necessarily want or need to change it, but I doubt this is well known. Making the policy clear would help, but I am not confident to change the wording significantly without approval. Let me just kick a thread on devel to see what others think as well. I'll cross-post a link here.

@alebastr it should be possible for people to process f31 branches now. I have created an issue in fedscm_admin to use PDC api, will implement it later this week.

So, I'm not sure where we are here? fedscm_admin doesn't enforce the current policy? or does it?

And do we want to change the policy? Or not?

Metadata Update from @kevin:
- Issue tagged with: meeting

a month ago

So we have a ticket for fedscm_admin to inform developers nicely about requesting branch for EOL release.

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

a month ago

Login to comment on this ticket.

Metadata
Boards 1
Dev Status: Backlog