Foundations Team Updates - Thursday 2024/06/06

Previous status: Foundations Team Updates - Thursday 2024/05/30
Next status: Foundations Team Updates - Thursday 2024/06/13


Banana PI BPI-F3

  • Setup board
  • Analyze boot flow
  • Determine deficiencies that would need to be resolved to support the board

Nezha D1





  • hiring
  • pulse planning
1 Like
  • Worked on making ppa-dev-tool snap play nice with LP credentials
  • Reviewed, and iterated on src:rustc-1.76 (still not uploaded, we keep finding new issues)
  • Spec review for versioned rustc and Rust dev snaps
  • Lots of needrestart triaging and discussions
  • Review of the openssl merge, with an ongoing discussion wrt the fate of the FIPS patches.
  • Patch Pilot
  • Interviews & THTs
1 Like


  • Fixed a bug which meant we lost test requests when we restarted the worker services
  • a bit of mp review for hyask
  • Amending some of my open Mp’s
  • replied to some rt’s - one about missing auto-sync images for oracular ppc64el, and one regarding some issues we’ve been having with bos03-arm64
  • Making an MP to restart our worker services whenever we update the cloud worker charm - something we previously had to do manually


  • Answering questions and providing documentation and helpful links for Jesus from the desktop team
  • Working on a PoC installer test using zapper and robot framework - it’s currently nearly done, just debugging the final steps! Hurrah!
  • Shared some info regarding zapper and robot framework with waveform over a quick call


  • Had a chat w Paride about charming our OpenQA installation once it’s done for both production and developer machines - hopefully making it much easier for devs to submit their own OpenQA test plans
  • Spoke with Sergio from the desktop team about placing their Core Desktop OpenQA tests on our infrastructure once it’s up and running

iso testing

  • The iso smoke default test for oracular desktop has been incredibly flaky recently - I’ve been spending time trying to fix this and will continue to do so
  • I retried said flaky test a bunch of times in order to get the amd64 oracular desktop image to promote from pending to current
1 Like


  • openssl 3.2:
    • Simon reviewed and has a comment about dropping patches related to simplifying maintenance of FIPS packages
    • unfortunately the patches don’t apply cleanly and will require some work to re-introduce
  • published a post about plans for upcoming ubuntu/openssl versions: Upcoming openssl versions in Ubuntu: 3.2, 3.3, 3.4,
  • cinder: drop python-oauth2client dependency in favor of python-googleauth (same purpose but newer and maintained), thanks @tsimonq2 for sponsoring
  • python-googleapi: also drop python-oauth2client for python-googleauth, again thanks @tsimonq2 for sponsoring
  • discussions around removing pydrive which also depends on python-oauth2client and has been replaced with pydrive2 (which also uses python-oauth2client however) but I still need to open a bug about its removal


1 Like


  • PR #2006 - I opened a PR to merge the netboot bug (LP: #2062988) fix to the stable branch after verifying it fixes the issue there
  • Investigation in LP: #2063331 - It turns out the policy outlined in that bug is already the intended behavior in subiquity, but there’s a bug in the way interfaces are marked automatically disabled (i.e. it works for manual disabled interfaces). Work to continue on this next week.
  • Updated the Apport MP #454563 for the updated subiquity hook after some good feedback (thanks @bdrung). I’m currently trying to write up a good SRU test plan for the proposed changes.
  • General bug triage


  • Last week’s proposed migration: still writing up the MIR for libsub-prototype-perl
  • I am on +1 maintenance this week, looking mostly at regressions due to the pytest 7 → 8 transition. I’ll send a more detailed report on the ubuntu-devel mailing list at the end of the week. Thanks @mwhudson for all the sponsoring this week.
1 Like


  • Investigated a regression reported for the dotnet-runtime-70 and published a fix (see canonical/dotnet-snap#14).
  • Started a discussion about snap symlinks on an issue reported to the dotnet-snap repo (see canonical/dotnet-snap#13).
  • Reported a snapcraft regression in the dotnet plugin on core22 bases (plugin not registered: ‘dotnet’, canonical/snapcraft#4825)
  • .NET 6.0.130 (May Update) sponsored to the unapproved queue on Mantic and Jammy (LP#2065300, thanks @enr0n).
  • Uploaded a Noble build of .NET 6.0.130 (May update) to the backports PPA from the backports-proposed PPA.
  • Updated the .NET snaps spec (FO-151) with items discussed during the Madrid Engineering Sprint regarding the delivery of .NET binaries using content snaps, with the added benefit of having automatic updates through the snap store, and the migration story from the legacy dotnet-sdk snap.
  • Started the implementation of content snap support in the .NET installer for the new .NET snaps. This is currently in the feat/new-structure branch of the dotnet-snap repo.
  • Attended weekly .NET partners sync meeting.
1 Like
  • Added some results to the dracut regression bug (LP: #2065180)
  • Verified Audio DAC+ board fix in the raspi kernel (LP: #2060240)
  • rpi-imager upstream kindly merged the change to use localectl
  • Provided a response on the armhf incompatibility thread (related to LP: #2038582)
  • Reviewed Core 24 addition to the rpi-imager JSON
  • Reviewed spec for initial user setup on pi images
  • Still playing with checkbox, zapper, and test-flinger for automating ISO tests
1 Like



1 Like



Started to analyze the packages that ship initramfs-tools hooks.


proposed migration

I looked at dracut blocking systemd 256~rc3-5ubuntu1. I need to merge dracut 102-1 to support systemd 256.

+1 maintenance

I was on +1 maintenance from 2024-05-30 until 2024-25-31. I looked at excuses... and worked backwards from the oldest to the newest entries. Remaining bits from +1 maintenance. Full report: +1 Maintenance Report

  • skorch / pytorch: skorch is waiting for pytorch. pytorch fails to build with llvm 18. I found a fix in the upstream git repository, but this is not enough to get pytorch building. I am leaving that as excercise for the next person looking at it.
  • rust-axum-core: update_excuse claims that rust-axum-core has no binaries on any arch, but 0.3.4-1 : rust-axum-core package : Ubuntu shows that librust-axum-core-dev was build on all archs. I asked in #ubuntu-release for an explanation.
  • python-mp-api: python-mp-api depends on python3-pymatgen (>= 2022.3.7~) and python3-emmet-core (>= 0.54.0~), but python3-pymatgen is not in the archive. This package is build by pymatgen. pymatgen 2024.1.27+dfsg1-7 was removed from Ubuntu with “Debian #1069219, ftbfs, no rdeps” as reason. I imported and requested updating to 2024.5.1 which adds Python 3.12 support.
  • libpsml: Imported FTBFS bug from Debian. This package needs xmlf90, but this package is neither in Debian nor in the NEW queue. But there is packaging done in
  • rust-imperative: Imported FTBFS bug from Debian. rust-stemmers is not in the archive nor in the Debian NEW queue.
  • spopt: spopt 0.5.0-4 fails to build from source, because it build-depends on python3-distutils. This build dependency is not needed. spopt does not use distutils directly. After removing the python3-distutils build-dependency, spopt fails to build because python3-spaghetti is missing. spaghetti was removed from Ubuntu because it was removed from testing due to I am leaving it to the Debian Science maintainers to address those issues.
  • umap-learn / python-pynndescent: umap-learn 0.5.4+dfsg-1 does not migrate, because it depends on python3-pynndescent but is not available. python-pynndescent 0.5.8-2 was removed with following reason: “Debian #1057598, ftbfs, scheduled for removal in testing, rdep umap-learn” I worked on python-pynndescent to get it updated. I picked up the work for 0.5.11-1 and fixed the test cases and released 0.5.11-1.

patch pilot

1 Like

OpenJDK / Java

  • Tried checkpoint/restore of a Java application inside a docker container and wrote up a blog-post.
  • Began work on the crac-jdk package. Preparing an upstream for -17 because the openjdk upstream patches sit on a 3-year old level. Needs re-basing considerations for ~120 patches. Most of my time is being spent on this task.
  • Did a comparison of openssl and BouncyCastle FIPS algorithms, plan to publish it formally as the next step.
  • Spent some time answering the Security team’s questions regarding the latest OpenJDK updates (which are now published).
1 Like


  • MP to fix and improve the admin page.
  • Toying with the removal of the flock for bos03-arm64, harder than expected because of mostly broken nova script. Resulting MR
  • Making measurements after a whole week-end A/B testing the flock in bos03-arm64, turns out it didn’t help.
  • Digging further on the arm64 VM taking usually more than 1000s to reach inside the testbed, and creating RT#164415 about this.
  • Finally opening the MP about the stats collection from the autopkgtest log files.


  • Investigated a couple of jobs failing in Jenkins, that pass successfully on my laptop. Turned out the base images were broken.


  • Putting it back on track after a reboot. Next ones should be go smoother.
1 Like



  • Review PR #457: added guide for contributing to the netplan documentation
  • Review (WIP) PR #475: Create
1 Like



  • working through UbuntuStudio image build regressions with move from “dvd” to “daily-live”


1 Like


  • Spec work for moving from v1 to v2 partitioning API
  • Working on adding v2 partitioning calls in the Ubuntu Server installer
  • Dropped all references to WSL/system-setup from the codebase.
  • Bug triaging
1 Like
  • NX support is still the big item
    • Further work on Ubuntu NX patches for GRUB
      • Mostly complete, but found a few bugs
      • Also found a new heap corruption bug :confused:
    • Further work on upstream NX patches for GRUB
    • Developed shim patch to build both NX and non-NX shims, made fb and mm NX_COMPAT.
      • Both shims work
      • mm fails to load under NX when launched from GRUB, needs testing
      • fb works
    • Developed shim-signed patch to add debconf and postinst based shim selection and install logic for the two shims
  • NVME-o-TCP grub-install support TBSD
    • ISO to be delivered from ogayot
1 Like
1 Like



  • Some investigations into frame pointer argument and a potential change for clang to record compilation switches into the binary
1 Like