Prepared Kinetic fix by cherry-picking the above fix. Uploaded, package currently waiting in kinetic-proposed due to a (likely) unrelated autopkgtest regression on systemd
Updated the autopkgtest charms and published them to charmhub
Also served as a test for permissions there
Tried to redeploy an armhf autopkgtest worker, deployment fail due to a systemd bug causing DNS resolution failures in containers
Committed a workaround for the issues with systemd-udevd by increasing the period which we’ll wait for the testbed to respond to ssh.
Running of cleanup-instances on the autopkgtest-cloud-worker units as the systemd-udevd kernel oops issue is leaving stray instances around.
Investigated what is causing the amd64 instances to boot slowly and causing the autopkgtest backlog. It seems to be an issue with the kernel as downgrading systemd did not resolve the issue. I reported bug 1989258 regarding the slow tasks timeout.
Modified the instance creation script so that the virtio_rng module is blacklisted allowing the instances to boot much more qucikly.
Tested an updated kernel for the kernel team which seems to resolve the issue.
Sent a PSA to ubuntu-devel regarding the issues with the autopkgtest environment.
Ubuntu
Added a force-badtest hint for nrepl-clojure as it dropped some arches that previously had autopkgtest results which had passed.
Declined a Launchpad invitation for ~ubuntu-core-dev to join ~ubuntucinnamon-dev and ~ubuntu-unity-devs (it should be ~motu per @vorlon). Accepted an invitation for ~motu to join ~farsi-fonts team.
Performed the SRU verification for an ubuntu-release-upgrader fix for bug 1987452 which is about a failure to upgrade from 20.04 or 20.10 if the latest version of apt is not installed.
Foundations team bug triage rotation x 2, once with @jawn-smith.
Let the Ubuntu Studio lead know about the possibility of adding a dist-upgrade quirk to ubuntu-release-upgrader to prevent upgrades from 20.04 to 22.04.
Submitted a pull request for ubuntu-server-triage changing who is considered a member of a team for TIL purposes.
Built a ubuntu-unity daily image per the request of the Ubuntu Unity lead.
Tested auto-upgrade-testing for @Paride and discovered a missing dependency so updated upstream with that missing dep.
Sorted out how to ssh to a test instance in lgw01 and then granted some kernel team members access to a system to experiment with the virtio_rng module.
Discussed with the CPC team the possibility of having minimal lxd images for armhf.
Merged my MP regarding blacklisting virtio_rng in Kinetic amd64 autopkgtest images.
Worked with @Paride to diagnose and resolve an issue with all the armhf workers going AWOL. Additionally, documented the incident and bugs discovered for future reference.
Committed @Paride’s cowboy’ed changes switching container providers from images: to ubuntu: .
Updated 5 armhf remote workers so that they have the latest version of autopkgtest-build-lxd, new cronjobs for creating updated armhf containers, and manually created new kinetic armhf containers.
Reported bug 1990319 regarding the autopkgtest code not getting updated on armhf remote workers.
Submitted an MP which allows us to build trusty lxd armhf containers again.
Ubuntu
Helped @enr0n get started on his Foundations bug triage day.
Reviewed and replied to multiple Foundations team triage reports.
Worked around by modifying the autopkgtest-build-lxd script used in the autopkgtest infra to disable (mask) snapd.seeded.service, umount the mount causing the Function not implemented issue and then purging snapd.
(Semi-)manual update of all the LXD armhf VMs to reflect the above changes without deleting and replacing them from scratch (mostly to avoid the “swirlix dance”: we don’t want VMs on swirlix).
Retrigger of some jobs failure due to the changes described above:
Collateral finding: the armhf cloud-init userdata in autopkgtest-clouds removed /snap/bin from PATH in /etc/environment. Brian has an MP up to fix this:
Fixed failing Ubuntu Desktop ISO testing jobs that prevented that Desktop daily ISOs from being proposed from -pending to -current
Underlying issue: test disk image not big enough, with consequent full partition and failures. Bumped the test disk image to 12GB (still less then the recommended minimum requisite)
General cleanup of venonat, with removal of lingering libvirt instances from failing ISO tests
Upgrade testing: uploaded auto-upgrade-testing 0.3.1 to Kinetic (bugfix upload)
Brought up (already known) issue with the build process of Desktop daily images (kinetic)
Discussed the above issue with @toabctl with the help of the newly introduced debugging messages, and found out the the problem was with a launchpadTrigger process lingering in background and not releasing a lock. (This should not happen as Jenkins should kill the entire process group of a job when it terminates, but somehow sometimes processes escape.)
cloud-init: preliminary work to enable integration testing on Oracle cloud.
Ubuntu Pro / ESM
Reviewed, merged and uploaded change to update-notifier for all the stable releases that updates ESM service name and description.
That is: uploaded update-notifier 3.168.16 (xenial), 3.192.1.13 (bionic), 3.192.30.12 (focal), 3.192.54.1 (jammy)
Issue: the changes in update-notifier 3.192.30.12 (focal) weren’t pushed to the packaging repository (this is a native package with per-release branches). Reconciled w.r.t the archive by cherry-picking from git-ubuntu import.
Issue: turns out there were more inconsistencies between the source package in the archive for Bionic and Focal and the upstream repository (= packaging repository, as the package is native). Thanks @lamoura and @ahasenack for sorting that out and reuploading.