Previous status: Foundations Team Updates - Thursday 2024/08/01
Next status: Foundations Team Updates - Thursday 2024/08/15
1 Like
subiquity
- Fixed regressions when handling debug shells on development branch. Regressions were introduced when handling the urwid 2.1 → 2.6 update. Fixes related to debug shell in non-async context by ogayot · Pull Request #2044 · canonical/subiquity · GitHub
- Added a way to use fresh storage information although no disks were initially detected. This scenario is common with network storage devices (e.g., NVMe/TCP) and previously required a restart of the installer. storage: add a refresh and back buttons when no disk are discovered by ogayot · Pull Request #2048 · canonical/subiquity · GitHub
- Opened guided: set _netdev when guided direct by ogayot · Pull Request #2040 · canonical/subiquity · GitHub as a draft PR to “properly” handle NVMe/TCP in guided scenario. Needs to be revisited.
- More NVMe/TCP work for the PoC. The work in progress is available at GitHub - canonical/nvme-tcp-poc
- Investigated missing call to
zfs-pool export
at the end of some installations. Needs more work but it seems to be caused by earlier failures preventing the cleanup code to be called. - Backports for 24.04.1
- Code review
distro
- Uploaded 3.4.0-2 : python-argcomplete package : Ubuntu to address failed autopkgtest ; and added the fix autopkgtest issue in my SRU for noble (not accepted yet) (thanks @bdrung for sponsoring both uploads!)
- Updated SRU for Bug #2073461 “[SRU] virt-install ignores --cloud-init when --loc...” : Bugs : virt-manager package : Ubuntu (need a sponsor )
1 Like
Go
- golang-defaults has been updated to 1.23. Working on the proposed migration for that.
- containerd needs patch to build with with Go 1.23. But due to the current ongoing Go libraries transition, containerd/1.6 FTBFS with -proposed. The patch is now integrated in Debian package with new version, will merge the package later.
- golang-gvisor-gvisor needs a new version to build with Go 1.23. Updated in Debian and synced to Ubuntu.
- Update Go 1.22.6 & 1.21.13 releases.
1 Like
imagecraft
- PR # 53 - Make imagecraft build an image
- Fixed several bugs to build an image.
- Also documented how to build an image at the same time.
- In review.
- PR # 54 - Add basic spread tests
- Done and in review.
- I am discussing a bug in craft-parts preventing the build from working on 20.04 (or in fact any build base too old to have git able to properly a specific clone command)
- PR #54 - Migrate to craft-application 3.X
- Needed to properly select the build provider (and specifically multipass)
- Removing some code (now in the libs) and adapting to new concepts (BuildPlanner, etc.). This kind of things is “safer” to do now that we have basic spread tests.
- Opened LP #2076320 because the automatic schedule build of the imagecraft snap is not working (probably because the recipe references a external repo)
- I will trigger rebuilds by hand for now until I know if the LP team plans on fixing this soon or not. If not, I will mirror the repo on LP but I would like to avoid that because it slows down the build of a new snap when pushing to main.
ubuntu-image
- Providing support to several users
- Reviewed, added spread tests and merged PR # 239 - snapcraft.yaml: Fix config installation. Thanks @tobhe!
- Investigated some more LP #2057885
- The workaround in place is not satisfying our users.
- We probably need to revert our change andset the gpg homedir in
/tmp
again.
- PR # 222 - Run spread tests on noble too
- Rebased and merged.
- PR # 240 - Add dosfstools (needed to have mkfs.vfat)
- Done and merged.
- PR # 241 - Clean /dev/, /run/, /sys/
- Done and in review.
Misc.
- internal security compliance trainings
1 Like
distro
- python-docutils/0.21.2+dfsg-2 proposed-migration: investigated ruby-github-markup blocking python-docutils, but a new Debian upload fixed the issue so just needed to wait for auto-sync and retried autopkgtests. Also retried autopkgtests for snakemake and got them to pass.
dotnet
- Implemented a notification mechanism using systemd-path that allows the strictly-confined content snaps (.NET Runtime and SDK snaps) to notify the .NET Installer, which sits in a classical snap, whenever they get updated (refreshed) so that the installer can update the .mount systemd units. These units contain version numbers in the path they bind-mount, which is the reason why every update needs a new .mount unit.
- Working on CI validation tests for the .NET content snaps (canonical/dotnet-content-snaps#6)
1 Like
Support for upcoming RISC-V board
- Contribute to vendor U-Boot fixing numerous issues both in vendor code and in upstream.
- Create a first live-installer image.
1 Like
- Found u-boot FTBFS issue was CFLAGS being passed to a cpp invocation during cross-compilation; worked with @xypron on a fix for upstream
- Uploaded u-boot merge of 2024.1+dfsg-5ubuntu1 (LP: #2075352)
- Confirmed work-around for LXD oracular container issue under focal’s 5.4 kernel (use 5.15 kernel, with options to force cgroups2)
- Enabled FIT images in u-boot (LP: #2054092)
- Sponsored patch for JH7110 card detection fix for @xypron (LP: #2069410)
- Provided some MIR advice on X13s settings
- Assessed an issue with py3clean in maintscripts (LP: #2075337); provided some support on fix, and analysis of affected packages (many thanks to @bdrung for further analysis too!)
- Ongoing work to SRU affected packages in universe as MOTU
1 Like
Distro
- I spent a lot of time on py3clean fails when using alternate character set: I wrote a Python script to check all packages in the archive. That script ran over the weekend on the noble package. The tests for the jammy packages are still running.
- lftp: Merge 4.9.2-3 from Debian
- pymatgen: Sync 2024.6.10+dfsg1-4 (the Ubuntu delta was merged)
- h5py: Merge 3.11.0-5 from Debian (for pymatgen sync)
- Merged amd64-microcode 3.20240116.2+nmu1
Apport
- apport: mark apport with py.typed marker
- Very productive Apport debugging session with @cpete
- Reviewed and merged recent-syslog: read stdout after process completion
- Reviewed and merged test: wait_for_gdb_sleeping_child_process
- Reviewed and merged tests: fix flaky tests waiting for sleep command
- test: Add more type hints
- test: put apport-retrace temp files into /var/tmp
- apt_dpkg: Map mirror to correct URL for the architecture
Sponsoring
- Sponsored Sync openmm 8.1.1+dfsg-1 (universe) from Debian unstable (main)
- Sponsored fix python-argcomplete autopkgtest fails with TERM=linux
- Sponsored Sync cpuid 20240409-1 (universe) from Debian unstable (main)
- Sponsored Sync ike-scan 1.9.5-2 (universe) from Debian unstable (main)
- Sponsored Sync mongo-c-driver 1.27.4-1 (universe) from Debian unstable (main)
proposed migration
- Uploaded initramfs-tools 0.142ubuntu31 to oracular with fixes for the build and for the autopkgest
+1 maintenance
I was on +1 maintenance last week. The full report can be found on ubuntu-devel: +1 maintenance report - 2024 week 31
patch pilot
- Sponsored Sync nbd 1:3.26.1-4 from Debian unstable for oracular
- Sponsored Sync tftp-hpa 5.2+20240610-1 (main) from Debian unstable (main)
- Unsubsribed ~ubuntu-sponsors from Sync wine-development 8.21~repack-1 from Debian sid
1 Like
- Faffing about with GRUB2 hwe packages for stables (still work to do)
- Fix for LP#2073634 [grub2] Enable suppression of /EndEntire message
- Oracular committed
- SRU TBD
- Sent upstream
- Wrote shim NX support FFe
- To be reviewed
- Fixed GRUB2 FTBFS
- GRUB2 2.12-5ubuntu2 submitted for signing
1 Like
Netplan
- Prepared the paperwork for an SRU to Noble (LP#2074197)
- Added the Tiobe TICS analyzer to our CI (PR#498)
- Fixed a bunch of compilation errors when
-Wconversion
is enabled (PR#496) - Code review for a community contribution (PR#495)
- Investigated a problem with custom umask that started to happen after the security fixes landed. The /run/system/network might end up being unreadable by networkd (PR#497)
1 Like
- glibc:
- 2.40 upload
- preparing a possible SRU for framepointer stuff on Noble
- Rust:
- review and upload of 1.78
- some more rounds of review on a spec
- sponsored valgrind upload for new upstream version
- zlib: looked into s390x FTBFS
1 Like
TCK on armhf
- Concluded investigation on the three non-conformances observed in TCK-17 and -21 on armhf
OpenJDK/CRaC
Java FIPS provider
- Major refactor in the context of moving to the JUNIT test framework
Misc
Adoptium Workgroup meeting
Netherlands visa application submission
1 Like
- subiquity
- PR: #2035 - drop discourse reference as support location per Supporting Users
- PR: #2036 - reference updates for source, shutdown, ssh
- started PR: #2039
- proposed-migration: python-dateutil
- report Bug #2074530 “FTBFS on i386 - build-dep python3-coverage not ava...” : Bugs : python-dateutil package : Ubuntu and investigate workarounds
- but ultimately is fine with migration-reference tests (thanks @bdmurray) so leave it at that
1 Like
Apport
- I spent most of my time this week looking at flaky Apport tests and fixing some of those
Subiquity
- PR: #2042 - Stable branch merge
Distro
1 Like
- GRUB
- Implemented string conversion for fdt properties for use in fdtdump command
- Implemented --node option for fdtdump to access nested properties
- Lenovo X13s enablement
- Ubuntu Image
1 Like
Documentation
Charmhub
- Set up a PoC repo + RTD build for single-sourcing data platform charmed DB docs
- Designed the single-sourcing logic for partial and complete variant consolidation
Subiquity
- Small lang. formatting fixes: PR #2043
Developer Guide
- Review of Rust how-to: PR #6
1 Like
hardware-installer-testing
- Refactored and tidied up a fair bit of the repo
- modified the jenkins jobs for the tests to make them production ready
- 2 pr’s to add boot steps for new machines in order to boot them from usb
testflinger
- a couple of small pr’s: one to fix a string formatting issue in some logging, another with a fix to some documentation
autopkgtest-cloud
- investigation into some issues we had with seed-new-release when opening up autopkgtest for oracular
- fixed said issue - MP
- fixed a traceback for one of our juju actions - MP
- investigating issues with bos03-s390x - proxy issues and issues ssh’ing to units - still ongoing
lpci
- continuing my work on adding deb822 support to lpci
+1 maintenance
- MP
force-badtest
-ingpython-oauth2client
- waiting on release team review
britney
- Iterating on a change to stop queue-ing unnecessary tests on i386 - still ongoing, see recent commits to britney
- wrote a script to parse and search britneys log files to help with the above - might put it in
ubuntu-archive-tools
if something similar doesn’t already exist
misc
- tested this bug r.e. xubuntu 22.04->24.04
- training request
1 Like
autopkgtest-cloud
- MP review for @uralt
- Some investigation for @ogayot’s issue with python-argcomplete and zsh
- Opened an RT to officially ask about nested virtualization on arm64 and s390x
- Tried to dump some metadata about the running test in their log files, but it proved harder than expected. I’ll have to discuss the way forward with the team.
- Cowboyed a fix for a user page 500 error reported by @mitchdz: the cowboy adds some logging in addition to catching the exception, so that we can understand why the exception is thrown in the first place, and make sure it’s not hiding a real issue. I’ll open the MP with the long term fix once I have more data on this, but so far, nothing appeared.
- Started working on FO194 about continuously re-running some autopkgtests (like tmpfails or fails ones). This is now in the first round of review.
- Worked on embedding the
autopkgtest-cloud
code with agit clone
instead of bundled in the charm, to ease the cowboys, updates, and testing branches. I’m currently blocked at the testing stage: my current charm seems blocked on some unrelated error around theapt
layer. I might have to do a bit more than just ajuju deploy
.
autopkgtest
- Preparing the SRU for autopkgtest/5.37ubuntu4
python-amqplib
- Investigated further the
python-amqplib
issue, turns out the issue is also present in Debian, so I’ve opened a bug
MIR retry
- Opened a MIR for
retry
, because it’s a new dependency ofautopkgtest
1 Like
autopkgtest-cloud
- Worked on adding a cache for users who are allowed to run tests, hopefully cutting down on timeouts
autopkgtest-cloud
- Finished work on UX improvements, will slowly be getting them merged in. Namely,
- All dependencies are properly verified before downloading anything
- Downloaded images are cached, and cached images are checked for freshness
- Check if required disk space is present before trying to resize image
qa-jenkins-jobs
- Added linting to the jenkins jobs, then worked on getting the linting to pass
1 Like