Patch Pilot Hand-off 24.10

Syncs:

Merges:

SRUs:

  • extruct - waiting in Noble UNAPPROVED

Unsubscribed:

  • fail2ban
    • High-traffic bug, I’d encourage the SRU team to be incredibly careful in reviewing this given the audience for the package.
  • gnome-initial-setup
    • Already sponsored and migrated into the release pocket for Oracular, but no debdiff is present for Noble (or the SRU format). Asked OP to resubscribe once that is done.
  • Marked fcgiwrap as approved, @costamagnagianfranco sponsored.
  • Unsubscribed sponsors from bug 2064187 (llvm-toolchain-18/noble/UNAPPROVED) - @costamagnagianfranco sponsored.
  • Pinged @jamespage about the cinder MP - removing OAuth2 support in favor of Google Auth support seems slightly questionable and I’d like a second set of eyes.
  • Bug 2059156 needed an upload to an orphaned package in Debian; uploaded. That being said, Noble sponsorship is waiting on an SRU description. Commented and unsubscribed.
  • Sponsored landscape-client.
  • Sponsored fstransform for Focal, Jammy, Mantic, and Noble. Oracular is already Fix Released. Thanks @sudipmuk!
  • Sponsored ppa-purge for Mantic and Noble, fixing ppa-purge.
  • Unsubscribed sponsors from the adsys SRU bug, looks like there is nothing else to sponsor.
1 Like
  • routes LP: #2069139 reviewed and commented
  • dotnet6 LP: #2065300 already sponsored, unsubscribed ubuntu-sponsors
  • mrcal LP: #2064959 nothing to sponsor (yet), proposed a patch to try to move things along

Apologies for my late report from Monday.

Handover

  • golang-github-containers-common interaction with AppArmor since Noble LP: #2040483: this looks like a significant bug for which quality contributions are being made but could really use some help from a pilot. However, then I noticed that a related MP is not appearing in the sponsorship queue. Figuring out why led me into a rabbit hole to fix it that took most of my shift (see below). I suggest that the next pilot prioritises piloting these contributions next if you could please.

Other work

  • cpu-checker switch from egrep to grep -E: unsuitable for Ubuntu; rejected with an explanation

  • os-prober LP: #530252: this looks like a good high quality contribution. Asked the contributor about the use case (this blocks me from reviewing it further), but also asked internally if we can prioritise review of it.

  • glusterfs fix for difficult-to-reproduce crash. No upstream release yet. Needs more detailed review and involvement of affected user in Test Plan for SRU QA. But already assigned to @bryce by server team triage, so I was going to coordinate with him but I didn’t manage to back to it (see below) and I see that he’s addressed it now. Thanks Bryce!

Rabbit holes

  1. It turns out that when I enabled support for git-ubuntu MPs I missed MPs targetted to the applied branches. This needs fixing, but the sponsorship queue takes forever to run which hurts development iteration, so I wrote some tests for the already quite testable function that needed amending and verified that the fix made the tests pass. I still need to adjust the codebase to run the tests on “package build”, etc.

  2. I fixed up my previous MP for ubuntu-sponsors-reporter which is another case where items are known to get lost from the sponsorship queue. But fixing it unveiled a rabbithole of a 1 vs. n problem. I’m halfway through resolving this.

I had to move on from the rabbit holes but I think they are important since otherwise we will leave contributors waiting without them realising that their contributions are missing from the queue. I expect I’ll look at them in lieu of my next shift.

LP: #2040483: AppArmor denies crun sending signals to containers (stop, kill). SPent a long time on this one, which packages need rebuilding, and also tested the fix. Turns out it’s a bad upgrade experience, thinking about an SRU, because just updating the package doesn’t reload the apparmor profile, which is embedded in the podman binary. One has to either reboot, or manually unload the containers-default-0.57.4 profile. If unloading, of course that will make all running containers suddenly unconfined, without a way to confine them again because there is no text file for the apparmor profile that could be given to the apparmor_parser tool. I left some suggestions in the bug on how we could perhaps handle an SRU, and sponsored the oracular package. On another note, for some reason I cannot nominate the golang-github-containers-common task for noble, even though it’s published in noble. No idea what is going on, but it’s definitely needed in noble.

  • LP: #2034648 – optee-os-s32 – read through history; still pending response to Löic’s review
  • python-ldappool merge – still pending response to Simon’s review
  • policykit-1-gnome merge – this is now being persued in salsa but is pending response to Jeremy’s comments
  • vitrage-tempest-plugin merge – awaiting response to Simon’s review (merge may not be required)
  • compton-conf merge – needs one minor clarification on the maintainer address but otherwise looks good
  • LP: #2066148 – criu – reviewed and left some comments on patch headers and a request for forwarding of one patch
  • LP: #2058622 – lenovo-wwan-unlock – reviewed and left comments about lintian errors that need addressing
  • LP: #2067125 – mariadb – added targetting and triggered requested autopkgtests

Well, that was an interesting shift. I ended up working on sosreport only.

I started looking at the first bug in the queue, which had lots of history and led me to the last bug in the queue :slight_smile:

After working my way through the long history of that, I sponsored it for Oracular, Noble, Mantic, Jammy & Focal, after some tiny fixes.

That finally allowed me to sponsor that tiny sosreport fix for Bionic from the first item, too:

Both sosreport items are now handled and Sponsors unsubscribed.

So long, and thanks to @arif-ali for the IRC assistance!

1 Like
  • LP: #2069060 - fixed changelog version and sponsored
  • LP: #2070171 gave a recommendation for a more long term fix
  • LP: #2070186 Created a PPA and tested the new autopkgtest, requested a more descriptive changelog

I bluntly went through the queue in chronological order to avoid getting things too stale, only skipping items that were last commented by a potential uploader.

I continued where @schopin stopped going from old to newer. I reached 2024-06-17. So leaving the green and white entries to the next one.

I continued from where @bdrung stopped, and I was able to do the following:

  • node-yarnpkg FTBFS fix: I did some adjustments myself and sponsored the upload.
  • python3.10 SRU: I felt we still need some test cases in the proposed patch before we move on. Added a comment stating so.
  • python-oslo.messaging SRU: The proposed changes look good but the version string seems weird to me. Asked the rationale behind that and also proposed a new version string which seems more sensible to me.
  • jsxgraph FTBFS fix: Disapproved this MP since the bug was fixed by the Debian maintainer and the fixed version was already sync’ed into Ubuntu.

Today I’ve spent my time making the sponsorship report better.

Unfortunately just running the script takes about an hour, so working on it is burdensome. I have multiple branches that are mostly ready but testing and iterating on minor polishing issues is extremely slow. While waiting, I started a performance branch that approximately doubles the report generation speed. But I’m still waiting for it to run to finish testing it!

Summary

Useful context:

  • The wsdd uploads look good, and looks like the patches can be applied at the same time, making it a single upload.
  • gobi-loader LP: #2071459 already sponsored, unsubscribed ubuntu-sponsors
  • nagios-plugins-contrib MP 468627 already fixed by a merge from Debian unstable, marked Disapprove
  • joyent-mdata-client MP 467487 reviewed and sponsored
  • python-cramjam MP 467617 reviewed and sponsored
  • cura-engine MP 468332 reviewed and sponsored
  • lshw MP 468665 reviewed and sponsored