The previous status is here: Foundations Team Updates - Thursday 29 Jun 2023
short week - one day off
- finished and merged all OEM bits related to meta-packages and kernel installation
- working on testing on certified hardware and fix any issue encountered
- learning about casper and how it does log persistence
- merged php-doctrine-cache (thanks @lucaskanashiro for sponsoring!) to unblock sqlite3 migration.
- armhf time_t investigation: batch 39
- More travel planning
- Uploaded r-cran-irkernel and r-cran-gnm, and managed to get r-base to migrate
- Found most of the arm64 autopkgtest runners were running
migration-reference/0tests for linux-nvidia-5.19 which run for about 3 to 6 hours and fail anyway. The same was happening on amd64, but the test only takes an hour to run there. Added linux-nvidia-5.19 to never_run and the arm64 queues started to move again
- Sponsored upload of dotnet7 for @dviererbe
- Sponsored syncs of openjdk-8, openjdk-19 and openjdk-20 for @vpa1977
- Found many s390x autopkgtests were “looping” (being put back in at the front of the queue after an infrastructure failure). The result was almost all of the runners were occupied running tests for packages that were unlikely to pass. I tried to mitigate this by temporarily adding some packages to
never_runand some tests have been running to completion
- Fix an issue with virtual packages (everything after one would be skipped): https://salsa.debian.org/vorlon/armhf-time_t/-/merge_requests/83
- Add .shellcheckrc after cleaning up corresponding warnings (shellcheck would have caught the issue fixed above): https://salsa.debian.org/vorlon/armhf-time_t/-/merge_requests/85
- Some work on the cdebconf merge, trying to do it the git-ubuntu way although it’s not currently a git-ubuntu package
- Work on crypto configuration testsuite, re-identified https://gitlab.com/gnutls/gnutls/-/issues/881
- Preparations to attend Kernel Recipes 2023
- Wrote “sanity” tests to assert availability/non-availability of crypto algorithms based on strict/non-strict FIPS modes
- LP team created (
- First LP MR, second upstream MR created
- mandatory HR trainings
- VISA documentation
- migration: (curl vs nss on armhf), https://ubuntu-archive-team.ubuntu.com/proposed-migration/update_excuses.html#nss
- migration: python-renderpm vs python-biopython, https://pad.lv/2025133
- merge: libnet-dns-perl, https://launchpad.net/ubuntu/+source/libnet-dns-perl/1.39-2ubuntu1
- merge: argus-client, https://launchpad.net/ubuntu/+source/argus-clients/1:22.214.171.124-6.2ubuntu1
- sync: findutils, https://launchpad.net/ubuntu/+source/findutils/4.9.0-5
- patch pilot
- Sponsored hg-git sync for @hjd after confirming the delta can indeed be dropped, https://pad.lv/2025558
- Sponsored a bunch of NEW/jammy oem-*-meta packages from the top of the sponsorship queue, needs AA review and needs to be moved to the corresponding package name, once accepted into the archive
- Sponsored sbuild merge, https://pad.lv/2025652
- Sponsored xfsprogs merge, https://pad.lv/2025656
- Some discussion about SRU versioning with @halves on #ubuntu-devel, https://wiki.ubuntu.com/SecurityTeam/UpdatePreparation#Update_the_packaging
- Asked for git-ubuntu style rebase on console-setup merge (to preserve git-ubuntu history), https://pad.lv/2025363
- address spec & manifesto review comments
- bug grooming
- How to use Debian cloud-images to try Netplan/networkd blog
- Debian networking discussions & investigations
- working on Netplan proposed-migration (vs systemd v253 & iproute 6.4), which already hits in Debian, soon to hit Ubuntu, too.
- DebConf paperwork
- Candidate interviews
- Pulse loading
Short week with holiday.
- package vendor packages into ppa
- adjust building preinstalled images for Nezha D1 and LicheeRV taking into account the change from kernel flavour allwinner to generic.
- Proposal kmod merge https://code.launchpad.net/~zhsj/ubuntu/+source/kmod/+git/kmod/+merge/446138
- Keep fighting with three packages carried from last week.
- Investigate autopkgtest regression for golang-defaults. These packages (vim-youcompleteme & ycmd) pass in local env and Debian, but keep failing on armhf in autopkgtest.ubuntu.com env.
- Start rebuilding the Go packages with golang-1.21 in PPA.
- Verify golang-1.20 SRU to jammy; coordinate golang-1.20 SRU to focal.
- libwww-mechanize-perl: one of the tests are deadlocking during autopkgtests. I was unable to reproduce the issue and proposed a patch to disable it to unblock the migration until we figure out what’s happening. LP: #2025645
- netcat-openbsd: Merge from Debian. (merged) LP: #2025666
sudo: Merge from Debian (pending sponsoring). Also investigated and proposed a fix for an autopkgtest that fails only on s390x. It seems that the ldap daemon is not ready yet when the ldapmodify command is called so I added an
sleepbetween them. LP: #2025655
- diffoscope: Merge from Debian (pending sponsoring). Also investigated a test that fails only on our armhf infrastructure (see below). LP: #2025887
- aapt2: Investigated why it’s causing one of diffoscope’s test to fail. Turns out it’s crashing when executed from an armhf container running on aarch64. The problem is described here LP: #2026151
- firebird3.0: Merge from Debian (pending sponsoring). LP: #2025754
- Investigated and proposed a fix for a problem reported by a Network Manager user. Network Manager can emit some special configuration for
cloned-mac-addressthat Netplan doesn’t support. LP: #2026230 and PR#376
- Did some refactoring in my “Wireguard private key flags” PR#371
- Spec/docs review
- Started working on the
- Finished desktop cloud-init blog post
- Worked on follow-up state-of-the-desktop round-up
- Merged sbuild 0.85.2 from Debian (LP: #2025652), thanks to @slyon for sponsorship
- Merged xfsprogs 6.3.0 from Debian (LP: #2025656), thanks to @slyon for (more!) sponsorship
- Attempted to lend a hand with the hugo sync (LP: #2025668), but ran into some build issues; awaiting some work upstream
- Reviewed some more
initramfs-toolschanges for @bdrung (autopkgtest on all archs \o/)
- Threw a spanner in the works of a CPC spec (sorry!)
- Worked on methods for expanding the boot partition on an existing (stock Ubuntu) Pi installation
- Debugging daily mantic images
- … including a day spent banging my head against the wall known as
- busybox merge uploaded
- (LP: #2026292) - with systemd 252, there is a conflict when moving between rsyslog and busybox-syslogd (in either direction) if you don’t purge the package between this change. systemd 253 resolves this.
- (LP: #2022102) - review https://github.com/canonical/cloud-init/pull/4192 and do a WIP branch for the subiquity-side of that work. When that cloud-init functionality + another part is on mantic, we can restore autoinstall via cloud-init on mantic.
- start work on zfs cache fix handling, found LP: #2026225 (subiquity/curtin kernel handling bug) during testing
- post-vacation review catchup
- Invalid PEP440 package version (LP: #1991606):
- Uploaded SRUs for dput and gpgme1.0, because the fixes for them are needed for the verification
- Verified SRUs for python-debian and distro-info
- Submitted Run autopkgtest on more architectures to Debian
and uploaded initramfs-tools 0.142ubuntu7 to mantic with that change
- Submitted test: Call qemu with -nodefaults to Debian
- Discussed making calling qemu-system-* easier
- Prepared dhclient replacement by dhcpcd (LP: #2024164). The upload needs to wait for the dhcpcd MIR in (LP: #2019191)
- Submitted Fix lintian complaints
- bug grooming
- Submitted Run autopkgtest on more architectures to Debian
- tzdata: Uploaded 2023c-7 to Debian unstable and merged to mantic
- bpftrace: Commented https://bugs.launchpad.net/ubuntu/+source/bpftrace/+bug/2023173 and unsubscribed ~ubuntu-sponsors
- thermald: Uploaded jammy SRU (https://bugs.launchpad.net/ubuntu/+source/thermald/+bug/1995606)
- v4l2loopback: Sponsored fix for mantic, lunar, and jammy (https://bugs.launchpad.net/ubuntu/+source/v4l2loopback/+bug/2025204)
- console-setup: Unsubscribed ~ubuntu-sponsors from https://bugs.launchpad.net/ubuntu/+source/console-setup/+bug/2025363
- netcat-openbsd: Sponsored merge 1.225-1 (https://bugs.launchpad.net/ubuntu/+source/netcat-openbsd/+bug/2025666)
Short week, with bank holiday on Tuesday and some sick leave last week
- SRU team shift on Friday
- some rebuilds for NBS packages.
- qgis was rebuilt in -proposed against new gdal but old Qt. Rebuilt again.
- processing removals from Debian (and then reverting some)
- time_t work: mostly merging in the work others did, and dividing up the next round of packages
- test on mantic/arm64 timed out
- re-triggered test passed
- new upstream release fails testsuite
- Heinrich Schuchardt (~xypron) already worked on it, but
I tried to fix the bug; work is still in Progress
- breaks on ubuntu but not on debian
- does not build for armhf
- there seems to be a patch that will be applied soon, which will solve this problem (see Debian Bug #1037439)
- failure seemed to be related to a transition, builds now fine
spend too much time in debian threads without a clue what is going on
P.S.: What witchcraft do you people use to get so much done in one Week?
- Open Issue #3544: 3rd party dependencies: Build against platform libraries instead of the embedded libraries
- attended weekly .NET Source-Build Partner Sync Meeting
- Worked on Issue 19: Overhaul Packaging Guide
- responded to IS Request of last week
- attended weekly meeting
- Submitted an MP resolving an issue with part of the autopkgtest-web reactive charm.
- Prevented port architecture tests from being run in bos01 due to networking issues there. Additionally, escalated an RT regarding the networking issues.
- Redeployed a lxd-armhf node which had a full /srv parition.
- Tested adding another team to the list of allowed teams which can trigger autopkgtests. The test successfully passed so I added canonical-foundations to the list of teams as they do a lot of +1 maintenance. Submitted an MP with the same change.
- Added rust-gtk3-macros on s390x, ppc64el, and arm64 to big_packages for @jbicha. Then requeued the tests which ended up all passing.
- Reviewing candidate applications.
- Administrative work.
- Patch pilot shift.
Benchmarking/optimization fun on memcmp. Turns out SIMD isn’t a performance panacea
- A few PRs reviewed and merged
- I got to write code this week, and quite a bit of it! Work to get APT deb822 support in apport yielded the following PRs: 192, 193, 194, 195
- Currently doing my patch pilot shift, sponsored @enr0n’s rsyslog merge and looking at his openssh one.
- Working on refactoring scripts that revolve around creating openstack instances and testing them.
- Attempting to fix issue with
seed-new-releasefailing after ~90 mins or so.
- Fixed issue with tests not being able to be requested via url in staging thanks to @sil2100
- Amended logging mechanism in
autopkgtest-webto make it compatible with apache and flask, now the logs actually go somewhere instead of into the ether
- confirmed a bug with the daily build (03/07/2023) of mantic desktop arm64 which when selecting the try ubuntu option, goes to a login screen and fails to login.
auto-upgrade-testing-specifications and qa-jenkins-jobs
- Created 2 mp’s which remove the timing specification in
qa-jenkins-jobsand instead specify a cron schedule for each individual test in
auto-upgrade-testing-specifications. We are trying to spread the tests out more evenly in an attempt to reduce the impact of venonat’s flakiness on the reliability of the auto upgrade tests.
Short week due to travels
- Continued investigation of
time_tmigration (batch 40 and 43)
- Working on re-activating
cargoMIR (LP #1993819)
- started work on openjdk-11-fips package
- sent patches for Java 21 compatibility for ant, plexus-compiler, maven-javadoc-plugin, javatools
- updated architectures patch for openjdk-22
- minor patches to enable sync from debian for openjdk-17,-19,-20
- Added merge proposal for LP: 2026215 preventing openjdk-8 migration
- raised last mp for batch 31
Away on PTO next week.