Foundations Team Updates - Thursday 11 May 2023

The previous status is here: Foundations Team Updates - Thursday 27 April 2023

(We didn’t have a status update last week, due to most people being at Canonical’s engineering sprint)

1 Like




  • engineering sprint week
  • take-home test reviews
  • candidate interviews
1 Like

Short week.

1 Like
  • Subiquity
    • PR: 1672 - fixes for snapd integration with the canary ISO
  • sprints, travel, 23.10 planning
  • greenhouse
1 Like

This week I am on +1 maintenance.


Test failed on the biggest test file. Hence I had ghdl added to the big_packages in autopkgtest-package-configs.

Building on armhf, s390x fails on purpose according to debian/changelog. The binary packages for architectures armhf, riscv64, s390x which do not build were deleted (LP #2019091).


Autopkgtests failed when trying to delete a missing file. I created an upstream patch which was merged and fixes the problem. umockdev - 0.17.17-1ubuntu1 (LP #2019122) contains the change.


Autopkgtest fails on armhf with stack overflow in pthread test. Upstream has a bug report for it and a patch 637662d92865 (“stress-pthread: use 64 bit tid_addr to fix stack clobbering on 32 bit platforms”). With the patch the stack overflow goes away. Instead we now get a timeout in the pthread test on armhf.


Postfix causes failures in autopkgtests which have postfix as a dependency (e.g. fwlogwatch). The post install procedure checks the domain name and rejects it if it ends with a dot. RFC 1034 defines that fully qualified domain names end with trailing dot (LP #2019195).


Autopkgtests fail I was not able to reproduce the failures when running the test locally. The autopkgtest log just supplies relative code addresses which I mapped to source lines (LP #2019219).


Created a fix in which was merged (LP #2018702)

1 Like





1 Like

Short week, with swap day after the engineering sprint.


  • removed NBS kernel binary packages from {jammy,kinetic}-{security,updates} as announced at
    • Did this now rather than improving the scripts first based on apw input, because ancientminister was running out of disk space due to size of the mirror and triggering IS alerts
    • dealt with fallout from a GKE customer who was relying on being able to install corresponding header packages from the archive for the host kernel into a container. Handed off to the CPC team.
  • Bootstrapped dotnet7 for jammy


  • Made progress on the migration of ancientminister to PS5. Can now rsync the mirror from ftpmaster.internal, so have done so rather than having to rsync all of this across from 3FP.
  • Worked on debugging why moving from kpartx to losetup causes loopback partition device nodes to all fail to be created. The answer is a difference in udev behavior on the host vs in a container. The solution here may be to change launchpad-buildd to just mount devtmpfs in the container used for livefs builds, but that is a change that would take a while to agree and roll out, so I am planning on rolling back the losetup change in the meantime so that we can build riscv64 images in mantic.


  • Continued analysis of armhf 64-bit time_t impact. Current stats:
    • 368 ABI-breaking libraries
    • 74 libraries requiring LFS handling in the transition
    • 2150 libraries with no ABI changes
    • 2004 libraries not yet analyzed (1200 with no reverse-build-deps)
1 Like


  • Working on .NET 8 Preview 3 bootstrap package for Mantic
  • Did a quick investigation on current state of .NET on RISC-V (on pause)
1 Like


  • Currently trying to merge the cargo source package into rustc source package as previously planned


  • Assistance with the investigation of armhf 64-bit time_t migration impact
1 Like

what is the use of canary ISO? I was unable to install from it.
Thank you

  • Short week (swap day after sprint, plus bank holiday for “man in hat sits on chair”)
  • Verified fix for DSI touchscreens (LP: #1970603)
  • Triggered autopkgtests for r-cran-dtplyr for @xypron
  • Sponsored umockdev for @xypron (LP: #2019122)
  • Finished batch 2 of the armhf time_t tests
  • Working on migrating netplan network-manager overrides to ubuntu-settings
1 Like

subiquity / curtin

  • Fixed multiples bugs caused by non-ASCII characters in GPT partition names
  • Reading about OEM ecosystem


  • uploaded os-prober 1.81ubuntu3 (thanks @dbungert for sponsoring!)
  • working on merge of sysvinit
  • sprint, days off
1 Like

Sprint last week, and short week this week.


  • Starting v253 merge and taking inventory of our patches and delta from Debian
  • Wrote a patch to expand the ConditionSecurity=audit check so that we can drop a downstream patch


  • Did the initial update for upgrades to mantic
  • Bug triage
1 Like


  • very short week
  • cleaning up (old) bugs for openssl and gnutls (in particular) on launchpad
1 Like


  • Began investigation into why the debvm autopkgtest is running continuously. It appears to be an issue with the triggering package which is a new version of mmdebstrap. Handed off working on this to @andersson123.
  • Review, merging of autopkgtest-package-configs changes adding ghdl to big_packages for @xypron. Then retried the tests for him.
  • Restarted jujud on an autopkgtest-web worker because the unit was in an error state.
  • Investigation into and hacking a way to protect the worker-lcy02.conf file which has information that would get overwritten.
  • Investigation into hardware rng in autopkgtest instances.
  • Worked with IS to investigate UEFI firmware type issues in PS5 (lcy02) versus scalingstack (lgw01).


  • Updated the meta-release file to include Mantic Minotaur which will be 23.10.
  • Planning / coordination regarding release dates for 2024.
  • Created rls-mm reports.
  • Discovered how to and updated the ubuntu-release KPI to display mantic and make it the default.


  • Interview.
  • Post sprints administrative work.
1 Like

Short week but many things to report since the last time I did it…


  • Helped investigating a bug on MySQL on armhf LP#2019203
  • Requested a sync for bsdmainutils PR#2018684


  • Many improvements on my error handling PR PR#334
  • Documentation improvements PR#337
  • Fixing a little math mistake PR#340
  • VRFs parsing fix PR#341
  • Improvements on error reporting PR#342
  • More improvements on error reporting PR#343
  • Plugging a memory leak PR#334
  • Critical assertion on glib PR#345
  • Documentation improvements PR#346
  • Fixing a call to systemd-escape PR#347
  • Another memory leak PR#348
  • Set NM as the default backend for nm-devices PR#349
  • One more memory leak PR#350
  • Create a Python binding for the load_keyfile function PR#351
  • Fixes for wireguard PR#352
  • Fixes for OVS PR#353
  • Fix for read-the-docs build failure PR#357
  • Fix for 8021x parsing PR#358
1 Like

Yes, it’s very early and kind of broken. I would suggest focusing more on the regular daily-live build, which will continue to get Desktop Installer improvements for Mantic but not be quite so broken. High on my personal list is Guided ZFS (not implemented yet, give it a few weeks), and other folks have improvements in progress as well.

1 Like

very short week due to travel and swap days:


  • provided April release to security team
  • work in progress: merge of openjdk-8 from Debian

armhf time_t tests:

  • worked through batch 8 of the tests
1 Like

Short week due to public holiday on Monday. Working on

  • readthedocs building jenkins job
  • got an upstream debian mr merged
  • Created a script to edit /usr/share/distro-info/ubuntu.csv for all machines in juju environment of bastion
  • Been linting autopkgtest-cloud repository to enable linting in CI
  • Discussion and research into adding more logging to autopkgtest-cloud
  • Today I’m going to do some SRU verification
1 Like