Previous week’s status can be found at Foundations Team Updates - Thursday 20 Jul 2023
Short week (Thu & Fri off)
- merge: sponsored dash for @dviererbe, https://pad.lv/2026730
- sync libwww-perl (cc @vorlon)
- patch pilot shift
- approved wireguard private-key-flags, https://github.com/canonical/netplan/pull/371
- fixed (Core22) renderer, https://github.com/canonical/netplan/pull/382
- confirmed Core22 renderer fix, https://pad.lv/2027584
- uploaded 0.106.1-7, containing the fix
- approve state PR, https://github.com/canonical/netplan/pull/379
- approve coverity fixes, https://github.com/canonical/netplan/pull/383
- package split, uploading 0.106.1-8~exp1 into experimental/NEW
- Python CFFI bindings WIP, https://github.com/canonical/netplan/pull/385
- Take-home test reviews
- Sprint sessions
- July security release, prepared packages for the Security Team. Currently only openjdk-17 RISCV build for focal still ongoing.
- Finished investigation and fix for JDK- 8312488
- working on a launcher application that configures java for the FIPS mode.
- PR: 1730 - finish tests on zfs action ordering (mount order matters), and fix handling of zfs canmount.
- PR: 1735, PR: 1743 (LP: #2022102) - fix autoinstall on mantic by consuming new flat-file info from cloud-init instead of poking around in cloud-init internals. fix for jammy is a no-change subiquity snap rebuild to get a newer cloud-init in the snap.
- PR: 1738, PR: 1744 - implement black & isort code formatting.
- PR: 1741 - speed up integration tests by skipping more commands. Was running
journalctl -b, only to throw away the content seconds later.
- PR: 1742 - finish work on swap partition creation for zfs guided.
- PR: 1745 (LP: #1961079) - enable ESP on raid on certain devices, and boot code on raid refactor.
- Working on LP: #2026225 - fix desktop and extra kernel included in install layers
- Working on zfs / zpool default properties to better match Ubiquity behavior
- MP: 447668 - enhance swap method to make it easier for subiquity to ask about projected swap sizes.
- proposed-migration: pillow vs pilkit (LP: #2028289) - fix a documented pillow deprecation in use by pilkit and upload
- file LP: #2028686 about snapcraft and git worktrees
- 7.0.109-0ubuntu2 & 6.0.120-0ubuntu2 (building in PPA):
- d/README.source: updated content
- replaced .NET 6 references in .NET 7 package (LP: #2009864)
- added support documentation
- added end of life process documentation
- general overhaul
- d/dotnet.sh.in: DOTNET_ROOT was unnecessarily set (LP: #2027620)
- d/t/essential-binaries-and-config-files-should-be-present: remove check if DOTNET_ROOT is set
- updated matching-pattern to only match 7.0.1XX releases
- d/watch file will fail now deliberately. See comment in d/watch for more information
- unify d/repack-dotnet-tarball.sh into d/build-dotnet-tarball.sh and updated command line interface
- d/README.source: updated content
- investigated solutions for package versioning problem with @mateus-morais
- attended weekly .NET Partners Sync
- attended mid-cycle sprint events or watched recordings (e.g., future of wiki.ubuntu.com discussion)
- attended MIR meeting
- refined JIRA tickets for mid-cycle review
- merged ubuntu-packaging-guide PR #22 – Add pre-commit checks
- Merge proposal to merge dash 0.5.12-6 from debian unstable got sponsored \0/
- LP: #1990621 prepared SRU to jammy
- Future boards will rely on U-Boot SPL loading main U-Boot from USB and NVMe file systems. I filed several patches to fix deficiencies in U-Boot SPL.
- I assisted field engineering on debugging a boot issue
- The SRU for OpenSBI 1.3 in Jammy, Lunar is now in proposed. Verification is complete.
Silicon vendor collaboration
- Work on concept
- review written interviews
- sprint sessions
- Investigated proposed migration regression reports for packages:
- pillow vs. django-simple-captcha (resolved)
- pcre2 (still investigating)
- Working on nvme-stas MIR:
- Creating an nvme-of test case for the package as recommended by the MIR team (in progress).
- Learning about
- Discussed a new versioning approach for the .NET packages with @dviererbe
- dotnet8-preview6 package ready but on hold until versioning issues are resolved
- Attended weekly .NET partners sync
- Refined JIRA cards for mid-cycle review
- After testing s390x instance creation, pinged IS regarding restarting neutron for s0lp4. RT 155441 is the ticket about getting this fixed for reals.
- Merged @andersson123’s MP which prevents duplicate test combinations from being submitted.
- Cowboy’ed a change in production to resolve an issue queueing autopkgtests after the autopkgtest-web charm was updated to not allow duplicate requests.
- Submitted an MP with the change I deployed in production.
- Reported bug 2028796 regarding the error queueing autopkgtests encountered by an Ubuntu Developer.
- Reported bug 2028801 regarding the rejection message saying the test was queued when it was in fact running.
- Stopped running Kinetic autopkgtests as it is EoL. Removed Kinetic autopkgtest images from swift.
- Submitted an RT regarding the Ubuntu release KPI collector not being able to communicate with any cdimage servers via rsync any more.
- Submitted a firewall change giving the Ubuntu KPI server access back to the cdimage mirror servers which then fixes our cdimage size and age KPIs.
- Reported an issue regarding the timers for the metrics.
- Submitted a PR which adds support for getting CD image size and age for Ubuntu Studio in the ubuntu-release metrics.
- Updated the cnf extractor for the EoL of Kinetic.
- Updated the appropriate meta-release files for Kinetic so that they use old-releases.u.c.
- Patch pilot shift.
- Reviewing candidate applications.
- Sprint sessions.
OpenJDK technical debt
- Investigated an intermittent test failure (seen only in the CI) and proposed to not skip the failing test.
- Autopkgtests - added sanity FIPS support tests (MR)
- Worked on two proposed migration bugs
- python-coverage vs. setuptools
- casper vs. systemd
Fixes came in from upstream debian for both. So, I didn’t have to do much. However, these two issues helped me with a nice practical introduction to proposed migrations and autopkgtests.
- Adoptium Steering Committee meeting
- OpenJDK FIPS code review
- More visa paperwork
- brainstormed about the dependency of the source model on other models
- reworked MP for dbus 1.14.8-2 to address comments. Sponsored by @schopin, thanks!
- Investigated dbus autopkgtest failures on i386 (was a good opportunity to learn about i386 tooling in Ubuntu). Turned out to be an infrastructure issue. Passed after a retry.
- retried python-test-stages on all arches to unblock tox 4.4 - which has now migrated
- zulucrypt - investigated autopkgtest failures on some types of test beds. Turned out that for some reason, zulucrypt attempts to
setuid(SUDO_UID)when the SUDO_UID variable is defined - which causes permission denied errors in the test-suite. Opened a bug report upstream to get the root cause addressed. Worked around the issue in autopkgtest by unsetting the SUDO_UID variable (sponsored by @slyon, thanks!) and forwarded to Debian. The package migrated. See https://launchpad.net/bugs/2023614
- python-pytest-trio: retried autopkgtest with additional triggers to allow it to migrate alongside python-trio. Migrated.
- onednn: investigated arm64 build failure with GCC-13. There is an existing bug report upstream and the issue was addressed in onednn 3.1. I backported the changes to onednn 2.7 in Ubuntu (this was sponsored by @tsimonq2, thanks!) and forwarded the delta to Debian. The package migrated. See https://launchpad.net/bugs/2028759
- internal meetings
- Working on deb822 migration for ubuntu-release-upgrader
- Got systemd migrated in mantic
- Started preparing for a systemd SRU for Jammy (and Lunar prerequisite SRU)
- Uploaded update-notifier to mantic to fix LP :#1998947
- Will soon SRU this to Jammy, but we will skip Lunar for this
- Worked on pillow vs pycoast proposed migration
Keep investigating packages that FTBFS with golang-1.21.
Done with most easy packages. Updated in Debian. Some are still waiting in the proposed-migration.
Retry several autopkgtests for prometheus Go libraries. Migrated.
Fix one flaky test in golang-github-juju-collections that blocks others packages to migrate.
- Preparing VISA documents.
This week seems to have been more about reviewing other people’s work than actually producing much myself
- Reviewed and uploaded Rust 1.69 then 1.70 for @liushuyu-011
- Reviewed and uploaded
s390-tools 2.28-0ubuntu1, then reviewed
0ubuntu2, for @Frank again
- Reviewed a few MPs during my Patch Pilot shift
- Investigation then subsequent fix of a testsuite problem while pairing with @bdrung
- Reviewed and merged a small bunch of PRs for the release
- Fix the armhf stripping fix (must be its 3rd iteration now?)
- Re-uploaded the focal and Jammy SRUs sans the memcpy SVE patches
- Clerical work to reorganize the way we work on the armhf time_t issue
- Interview and take-home tests
- Opened, commented, and/or added patches to following bug reports to decrease initramfs size and to make it faster:
- apport: Uploaded 2.27.0-0ubuntu1 to mantic
- Wrote a bug pattern for full /boot partition
- Discussed additional needed kernel modules in SRU for Tegra system
- Bug grooming (from > 800 a few weeks ago down to 755). This includes marking failing /usr/share/initramfs-tools/hooks/casper-memdisk as duplicates of bug #1658352
- Upload initramfs-tools 0.142ubuntu8 to mantic
- Send Update on reducing initramfs size and speed up to #ubuntu-devel mailing list
- lintian: Exclude not-binnmuable-* tags in ubuntu/main profile
- Upload nbconvert 6.5.3-4 to unstable for @schopin
- Reviewed and merged https://code.launchpad.net/~dbungert/ubuntu-sponsoring/+git/ubuntu-sponsoring/+merge/446529
- I spent my patch pilot shift mostly with cleaning up the https://code.launchpad.net/ubuntu-sponsoring code
@schopin and I had our bi-weekly Apport meeting. We took the opportunity to troubleshoot a hanging test case when
/etc/apt/sources.listis missing and found the culprit: tests: ui: fix hangup on test failure
- feat: Import translation updates from Ubuntu 23.10 (mantic)
- apt_dpkg: Drop rootdir="/" from apt.Cache
- Release apport 2.27.0
Focused on specifications and talks due to both a scheduled meeting and me and others being absent during August.
- changed the file layout of the crypto-configuration testsuite to match what would be installed on system
- started integrating nginx in the scheme as a test and demo
- archeology around gnutls and openssl configurations and their changes over time both upstream and in Ubuntu
- prepared a talk on crypto-configuration and almost finished its first specification (trying to fully finish it by EOW)
- will be on vacation most of August
- Slow week, medical issues
- Working on udev rule replacements; libcamera relatively simple, GPIO/I2C/SPI simple, UARTs proving a bit harder
- Reviewed @bdrung’s changes for initramfs-tools, and did some timings on Pi 4 and Pi Zero 2
- Finished u-boot merge with some extra patches from @xypron (LP: #2027789)
- Still writing second article on the state of the various desktop flavours on the Pi (if anyone’s wondering about the delay I wound up testing several I hadn’t anticipated, and then redoing everything due to some Wayland related findings)
- Pi meetings
grub 2.12 rebasing fun. Lots of rebasing fun.
- Learning about and making changes to Grafana dashboards - both for daily image metrics and autopkgtest-cloud.
- Released and tested new charms in autopkgtest-cloud - first since May. Tested in staging and prod and made some amendments based on reported errors from users.
- Fixing and working on the new usage of the
autopkgtest-requestorsteam implementation in
- Polished an MP for
utahwhich fixes an issue we’ve been having in venonat - desktop vm’s weren’t getting destroyed and undefined properly, leaving the vm in a
shut offstate with it’s storage staying on the disk and not getting deleted
- Added a new suite of HTTP exceptions for
autopkgtest-webin an effort to have more uniform responses to the user.
- Wrote some documentation for contributing openqa tests and running locally
- Added a timeout argument for a setup script in
autopkgtestwhich can optionally reduce the time waiting for a setup script to run
- Added functionality in
autopkgtest-cloudwhich maps allowed releases and architectures e.g. trusty is in ESM so we only want to run tests on amd64 for it.