Foundations Team Updates - Thursday 17 Aug 2023

Previous week’s status can be found at Foundations Team Updates - Thursday 10 Aug 2023

1 Like




  • Take-home test reviews
  • Candidate interviews
  • Pulse grooming
  • Bug grooming

2ish week edition

  • subiquity
    • implement zsys compatible dataset creation. VM testing and PR pending.
    • write a spec on handling of invalid cloud-config when attempting to do autoinstall.
    • help with 22.04.3 release and last minute fix for LP: #2013201, partition edit issue
    • Investigate issue with 22.04.3 and using autoinstall. When doing nocloud URL + hostname or URL + instanceid, the autoinstall data is not visible to Subiquity. Release noted this issue, root cause already fixed in cloud-init.
  • livecd-rootfs
    • MP: #449215 (LP: #2031027)- several fixes for Desktop ISO construction - fix language layers, add a required package. Merged and uploaded.
    • Started some launchpadlib-based hacks around better testing of livecd-rootfs changes. Was able to test above change locally the above Desktop ISO change prior to upload, but it requires a lot of knowledge on how the ISO is constructed. Allowed me to see the next issue, LP: #2031576 before daily builds were ready.
    • MP: #449308 - fix a ssh-keygen hang when running autopkgtest locally. Pending upload, would like a CPC ack as it’s their hook.
    • Investigate LP: #2028862, the hung startup issue with the symptom of snapd.seeded.service waiting forever, and if it would affect Jammy 22.04.4. The answer is yes (when cloud-init 23.3+ is SRUed). Proven with a quick backport of cloud-init 23.3 + inject into current live-server jammy daily. Will need a SRU of the change from livecd-rootfs.
  • util-linux: merge newer source package from Debian and apply a small test skip to address running the make check suite on a Focal kernel (LP: #2030793)
  • kiwi: rebase and upload (twice)
  • refpolicy, pilkit, cura-engine: delta issues fixed in Debian, syncpackage
  • greenhouse
  • participate in cloud-init summit
  • short week with bank holiday on Tuesday
  • will be out starting tonight - coming back on Aug 29th


  • focused on support of LUKS recovery key:
    • opened PR 1766 for subiquity
    • opened MP 449362 for curtin (adjustments would likely be needed for s390x where we use zkey)
    • manual testing in VMs is green. The recovery key is made available in the live session and is optionally (for autoinstalls only) copied to the target system under /var/log/installer.
  • testing of OEM implementation on certified hardware just started
  • Short week, national holiday
  • Merged mdadm from Debian unstable
  • Requested removal of r-bioc-isoformswitchanalyzer (LP: #2031293) and uploaded r-bioc-decoupler and r-cran-biocmanager to get r-bioc-* (170 packages) to migrate
  • armhf time_t full shift - week 33

Very short week due to bank holiday + day off.

  • New glibc upload last Friday to fix broken Breaks: relationship
  • a bit of -proposed migration work on glibc (although the heavy lifting has been done by @ginggs )
  • s390-tools: the new version 2.29 has some brand new Rust code, so I did the packaging work for that, and reviewed @frank’s work on the rest of the package before uploading. I also worked on the re-MIR for s390-tools that was triggered by that new addition (still a draft, not yet published).
  • aad-auth: reviewed & sponsored the 0.5 release for @denison-barbosa since it should make its autopkgtests happy with the new glibc :slight_smile:.
    • played a bit with rustix to try to get rid of linux-raw-sys entirely in aad-auth. No joy there, which is a shame as I would have done it in s390-tools as well.

Silicon vendor collaboration

  • Review and suggest U-Boot patches
  • Prepare installer image

Ubuntu summit

  • Prepare slides

Jira Planning

  • Update Jira cards relating to 24.04 cycle

Short week due to travels.




+1 Maintaince

  • go-exploitdb
    Blocks the migration of golang-gopkg-cheggaaa-pb.v1.
    Upload 0.0~git20181130.7c961e7-3 and request sync LP: #2031610
  • gost
    Blocks the migration of golang-gopkg-cheggaaa-pb.v1.
    Upload 0.1.2-2 and request sync LP: #2031609
  • golang-github-labstack-echo
    autopkgtest fails due to http_proxy env and png file in testdata is mangled.
    Upload 4.11.1-2
  • golang-github-form3tech-oss-jwt-go
    • Upload goval-dictionary and go-cve-dictionary to get rid of
      golang-github-labstack-echo.{v2,v3} and golang-github-dgrijalva-jwt-go
    • File RM request for golang-github-labstack-echo.{v2, v3}
      Debian Bug #1049945, #1049946
  • golang-github-mdlayher-socket
    Test only fails on launchpad. Narrow down to the kernel version from focal.
    Still checking the code.
  • golang-github-protonmail-go-crypto
    The error message says “out of memory”. But I didn’t observe such on a
    VM with 500M memory. Still working on it.


  • Rebased netplan diff after the new Python bindings with CFFI was merged
  • Coverity static analysis job is working. It will run once a week every Monday. The Netplan project can be accessed here
  • Fixed a couple of issues found by coverity PR#393
  • Changed the Coverity workflow to run only on our repository PR#392
  • Added support for VRFs in our keyfile parser PR#398
  • Added an iterator to access the list of routes from netdefs PR#397
  • Added an iterator to access DNS configuration from netdefs and a getter for the macaddress PR#395
  • Added getters for dhcp4 and dhcp6 options PR#394
  • Refactoring a few things in preparation for netplan diff. * Unfortunately netplan diff will not be ready for netplan 0.107 but all the support code was merged. PR#399
  • Support for veths was merged PR#368
  • Started investigating LP#2031421. NM is failing to add one of the routes defined for a VRF interface.


  • Investigated some oddities the Ubuntu QA team encountered when building the autopkgtest-web charm.


  • Tested a daily build of the canary image and encountered a crash which I reported as bug 2031027. (I was actually confirming an issue for @andersson123.)
  • Rebuilt (twice because I was too impatient the first time) the mantic-desktop-canary-amd64 image after livecd-rootfs migrated. Then performed an install test using TPM advanced option which failed to boot which I reported as bug 2031576.
  • Recreated a bug in software-properties which affects Ubuntu 22.04. I then uploaded the fix for SRU to jammy.
  • Modified crichton, a bug bot, to no longer subscribe ubuntu-sponsors to oem-* packages.
  • Tested a release upgrade from Lunar to Mantic given a failure seen in our automatic upgrade testing. The upgrade also had a failure for which I opened bug 2031420.


  • Conducted an interview with a candidate.
  • Reviewing candidate applications.
  • Patch pilot shift.
  • Submitted an RT regarding adding andersson123 to stg-proposed-migration on foundations-bastion-ps5.

Short week, on vacation last week


  • NEW processing for mantic, zeroed out the queue
  • process various removal requests, including a number related to +1 maintenance
  • catch up on Debian removals, filing a few bugs / MPs for reverse-dependencies
  • fielded a couple of i386 whitelist requests


  • Fixed a misbuild in petsc, caught by autopkgtests. Punted petsc4py autopkgtest failures to +1 maintenance.
  • Reviewed the various time_t quirk PRs that came in while I was away
  • livecd-rootfs MP review for canary image
  • Raised livecd-rootfs MP to clean up Desktop image squashfs layers
  • Reviewing the state of needrestart, canonical-livepatch, unattended-upgrades, update-notifier integration


  • Candidate interview for Commercial Systems


  • initramfs-tools:
  • dhcpcd: Sponsored dhcpcd 1:10.0.2-3ubuntu2 for Tobias Heider to fix strlcpy overflow in psp_ifname
  • dracut:
  • ubuntu-meta: Replace isc-dhcp-client by dhcpcd-base in minimal (LP: #2019722)





  • Review of an MR
  • Started spending time on learning and prototyping an openssl-based FIPS solution


  • Did a (late) talk submission to DebConf 23 named “Java and FIPS”
  • Public holiday on Tuesday
1 Like

Trying to install and boot windows from chainloader, explaining boot stuff, boot discussions. Surviving the high humidity somehow.



  • Created bootstrap packages for .NET 8 preview 7 for arm64 and amd64.
  • On Friday, the team decided to save the effort of bootstrapping preview7 due to a re-bootstrap being required when RC1 lands. Thus, the effort for a .NET 8 release is on hold until RC1 (september).
  • Attended the weekly .NET Partners Sync meeting
    • Discussed the possibility of not including the entire LLVM source code in the release tarball of .NET to avoid having vendored code built-in. Further discussions on Github.



short week with public holiday




  • raised issue for jexec fix.
  • discussed upstreaming –no-as-needed flag - will raise an issue and submit a PR
  • raised merge requests on salsa for javadoc search fix, patch refreshes and updated exclusion list in d/copyright since we bundle libcms


  • made a build system fix, change submitted to Debian


  • new upstream release ready for review. We need it to continue publishing openjdk-22 previews


  • a cleanup MR to enable build-time tests and align paths with specification


  • prepared a ppa build of openjdk 22+8 to evaluate upstream change.
    We might need to rebuild Renaissance test suite with openjdk-22 in order to run ML benchmarks where this patch promises good improvements.
1 Like