Foundations Team Updates - Thursday 14 April 2022

Foundations
1

Hello to all our Foundations fans (#fandations). This is the thread for our updates for the preceding week.

Last week’s update is available here: Foundations Team Updates - Thursday 7 April 2022

1 Like
2
  • subiquity
    • roadmap planning for 22.10
    • Jammy / 22.04.1 release prep
    • LP: #1967324 - Chat with @patriciadomin to learn the netboot process for ppc64el machines. Reproduce the failure a few times. Investigating.
    • PR: 1255 - Implement suggestions for fix to autoinstall / early-commands.
    • PR: 1261, PR: 1270 - Fix a crash that could possibly be seen when the server process restarts during autoinstall with interactive-sections, and merge to Jammy.
    • PR: 1262 - Merge RISC-V fixes for Jammy
    • PR: 1263 / LP: #1968160 - Add a filesystem reset to original state when going backwards from the Storage configuration screen to the Guided Storage screen. Also discussing a longer-term UX for this.
    • PR: 1264, PR: 1265 - Match the update to Ubiquity to the suggested boot size for a minimum of 1.75GiB, and merge to Jammy.
    • PR: 1267 - Draft PR exposing the resize support in curtin.
    • PR: 1269 - Enable partition deletion without having to format the full disk.
    • Review PR: 1257, PR: 1260, PR: 1266, PR: 1268, PR: 1271
  • curtin
    • MP: 417829 - initial resize partition support - address feedback and merge.
    • MP: 419137 - create Jammy stable branch, bring in RISC-V fixes.
2 Likes
3

NBS
continuing to keep NBS at zero following successfully clearing the list last Tuesday.

  • required syncs of astap, astap-cli following the update of lazarus with new package names

Proposed-migration

  • no-change rebuild of rnp, stuck in proposed, against current botan soname
  • disabled LTO in lorene, fixing the gyoto FTBFS on ppc64el and ultimately allowing us to remove python3.9.
  • synced new version of hyperspy now that python-sparse is fixed
  • tried to resolve stuck/uninstallable gcc-defaults-mipsen by syncing gcc-11-cross-mipsen, but this package is itself broken. Didn’t try further.
  • uploaded libjpeg8-empty to drop libjpeg-progs and let this package migrate
  • noticed that libpandoc-wrapper-perl autopkgtests are now passing, so dropped the block-proposed bug to let this migrate
  • libargs: pulled in the new upstream version of catch.hpp from the Catch2 project to fix build failure with glibc >= 2.34 (https://bugs.debian.org/1009611)
  • django-uwsgi: new version has failing autopkgtests in Debian and Ubuntu. Package not in Debian testing. Removed this version from -proposed (the version in the release pocket is apparently fine).
  • backupchecker: FTBFS (in -proposed and release) with python3.10 and removed from Debian testing, so removed it from jammy also.
  • adapterremoval: just like libargs, pulled in the new upstream version of catch.hpp from the Catch2 project to fix build failure with glibc >= 2.34 (https://bugs.debian.org/1009616)
  • wings3d: FTBFS bug report is marked as fixed in Debian, so synced the new upstream version from unstable.
  • wims: FTBFS in Debian and removed from testing, so removed from jammy along with its reverse-dependency wims-moodle.
  • rust-kmon: build-depends on removed rust-tui. Removed, added to extra-removals.
  • ruby-parallel: autopkgtest failure not seen in Debian. The actual test failures seem to have to do with timing. Retried the autopkgtests to see if they succeed when the infrastructure is calmer, but won’t invest more time in this.
  • ruby-diva: not migrating because it makes mikutter uninstallable. Debian removed mikutter from testing to let ruby-diva migrate, so doing the same (https://bugs.debian.org/997819).
  • racket-mode: autopkgtest failures on ARM not seen in Debian. Checked whether a no-change rebuild of emacs would help. It did not.
  • panicparse: FTBFS in Debian and Ubuntu, removed from Debian testing, now removed from jammy.
  • rust-futures: blocked because several packages depend on an obsolete package name. These have been removed from Debian testing for this issue, so removing from jammy as well. This unblocked a good number of other rust packages.
  • feedparser: blocked by an autopkgtest failure of rss2email. Retriggered with --all-proposed and it passed.
  • ck: upstream build system builds for wrong CPU target on armhf. Fixed and sent upstream to Debian.
  • libvbz-hdf-plugin: another catch2 vs glibc fix.
  • rust-gdk-pixbuf: autopkgtest was failing on !armhf with an OOM error. Added to big_packages list and it passed.

Archive admin

  • Resurrect packages that were removed because they depended on numba, which is now back
  • Rebuilt packages referencing obsolete exfat-utils package, allowing its removal from Ubuntu.
  • Zeroed out priority-mismatches after reviewing the libraries that remained. Raised a question on ubuntu devel about multiple json parsing libraries in the default seed; no one has responded yet.
  • Followed through on remaining MIRs for the cycle, driving component-mismatches to zero
  • Uploaded fwupd-signed to use correct Built-Using field, so that fwupd-efi source is kept in main where it belongs
  • Miscellaneous low-touch NEW processing of syncs from Debian
  • NEW processing of rpiboot, now in multiverse

Release

  • Miscellaneous review of FFes and uploads in the unapproved queue
  • Clean up i386 seed overrides in the code
  • Remove all i386 binaries which are no longer whitelisted
  • Check that there are no packages in extra-removals that need resuscitated for jammy
  • Land fix for bug #1946778 on all archs, so that it’s clearer we’re booting from an install image.
2 Likes
4
  • subiquity and curtin reviews
  • lots of roadmap planning
2 Likes
5
  • Trying to demote ltrace. ltrace broken
  • Submitted talk for DebConf22, prepared registration
  • Updated nullboot dependencies and setup code analysis (both GitHub CodeQL and lgtm) and dependabot; released 0.3.0 to jammy
  • I will be out of office Thu-Mon (inclusive)
1 Like
6

Short week due to travel

  • ubuntu-image:
  • Raspberry Pi:
    • Made some packaging edits for the Pimoroni Unicorn hats
    • Current working version exists in a PPA
    • Tested, reviewed and sponsored rpi-imager for @waveform
  • Golang
    • Pulled in new version 1.18.1 and 1.17.9 and pushed to Debian salsa
    • Uploaded to Ubuntu as well
  • Distro Work
    • Continued to investigate vim FTBFS on ppc64el
    • Not a memory issue, not an LTO issue, not an optimization issue
    • Builds fine in an schroot on POWER9
    • Added additional log statements to troubleshoot, waiting on PPA build
1 Like
7

Ubuntu

  • upgraded a laptop from Impish to Jammy and discovered / reported a screen lock bug
  • reviewed and merged ubuntu-release-upgrader fixes for @enr0n the screen lock bug also sponsored the jammy upload for this
  • uploaded apport disabling crash reporting to Launchpad for Jammy
  • determined that a couple of bolt regressions were false positives and overrode them
  • checked on the need to disable kerneloops uploads to Launchpad and submitted an MP removing it from the checklist
  • helped @bdrung use the Error Tracker and investigate an apport crash
  • reviewed, solicited additional feedback on, and uploaded ubiquity for @bdrung which fixes a long standing keyboard bug
  • created an impish branch of ubuntu-release-upgrader as this was missed
  • communicated with the Landscape team regarding subscribing to the landscape-client package in Launchpad
  • submitted an ubuntu-archive-tools MP switching from landscape to landscape-bugs
  • Reviews of packages in the unapproved queue for Jammy
  • SRU vanguard shift

Error Tracker

  • meeting with IS regarding the cassandra server move
  • investigation into the removal of old snap crashes in the Error Tracker. Currently, they are added with a TTL of 30 days but that didn’t used to be the case and while some of the vintage snap crash data was removed it seems multiple days, months, years(?) were missed. So I submitted an RT to re-remove the snap data.

Short week as I was on holiday last week.

1 Like
8

flash-kernel

GRUB

  • analyze failure to install Ubuntu Jammy via PXE (LP: #1967562): identified where network is disabled. This needs more discussion with upstream.
  • patch to disable Load File2 protocol on ARM as workaround

RISC-V installer image

  • Retesting

PolarFire SoC Icicle Kit

  • Investigating vendor software

RISC-V roadmap

  • sizing items
1 Like
9
  • Subiquity
    • Opened PR 1266 to address race condition in LP 1968161 [Block device probing screen sometimes not refreshing automatically]. We can do better to address the root cause (i.e., ensuring strict ownership of the display) but this is enough to address the race condition on the storage screen.
    • As part of PR 1252 to redesign third-party drivers screens, have been fighting with integration tests not playing answers for the Source controller. This needs to be addressed before this PR can be merged.
    • Reviewed various PRs.
    • Reported and investigated LP #1968729 [Cancelling HTTP calls from client side cause server side failure]. Opened PR 1272 to address the crash for third-party drivers. Now looking at similar but more complex failure in storage controller.
  • Misc
    • Reworked patch for ltrace FTBFS on ppc64el (LP #1967518). Thanks to @xnox for sponsoring!
    • Will be off Friday 15th and Monday 18th.
1 Like
10

RISC-V live-installer: last fixes and tests on the Unmatched board
VisionFive: Test plan + investigating GPIO/poweroff failures
Nezha D1: Rebased Samuel Holland latest 5.18 branch onto 5.17 ubuntu unstable branch
Packages: indicator-keyboard LP #1968333, rnp uninstallable investigation rebuilt by vorlon
Kernel: Start implementing proper fix for LP #1967130
Roadmap 22.10

1 Like
11

Netplan

OpenSSL

Rust

  • Started working on Rust 1.59 as it is needed for the upcoming Firefox release.
1 Like
12

General

  • SRU reviews and releases
  • Helped sync gazebo from Debian per community request
  • Various discussions
    • Had a really nice talk with Michael!
  • Reviewed Ben’s ubiquity fix for the elusive invalid-keyboard-layout bug - good catch!
  • Started looking into VVAS packaging for Xilinx, but ultimately delegated this to Benjamin
  • Started working on getting some other robotics-related packages sponsored - but still not there yet

Release

  • Preparations for the incoming Release Sprint
  • Queue reviews for 22.04
  • Creating .1 and .2 milestones on Launchpad
  • Work on pre-release tasks from the checklist
    • Refreshed ubiquity translations, checked up on others. Pinged mvo for DDTP translation updates
    • Updated hwe-support-status in update-manager for 22.04
    • Got in contact with the web team
  • Last-minute FFe reviews (a bit too many for my taste!)
  • Reviewed some documentation regarding RISC-V installer usage

ubuntu-image

  • More work on the redesign spec, lots of cleanup of the document
  • Reviewed and iterated William’s on PR for filesystem edge-case handling
    • We have a good idea about what kind of testing we’d like for such changes now
  • Discussed ubuntu-image 2.2 autopkgtest failure situation with William

ubuntu-cdimage/debian-cd

  • Experimented with the created WSL actions credentials for running builds from ancientminister using request-rebuild - looking good!
  • Iterated on a few small changes that were needed in build-remote from the WSL repository
    • Almost done with a PR for upstream
  • Moved some of the temporary boot-asset management from the debian-cd codebase to cd-boot-images-riscv64. I’m not entirely happy with that, as I think the cd-boot-images-* workflow is slow and unnecessary.
  • Moved the riscv64 subiquity installer image builds to default_arches, meaning now live-server builds will take longer as we always wait for all the arches to build successfully before cdimage can publish those - and the riscv64 installer images take ~6 hours to build.

Ubuntu Core

  • Investigated the mystery of no new core20 builds in the last days (IMPOSSIBLE!)
    • CI Automation scripting was failing because, for unknown reasons, the state-file we stored in SWIFT seemed corrupted and included a plain-text HTTP header at the beginning
    • Added some debugging to the CI and, just to be sure, added handling of such cases, automatically trying to fix the statefile by removing the unneeded cruft
  • Promoted new core20 test snaps to a branch for some partners. Had to deal with those entering the review queue - so poking the security team a bit
  • Reviewed the UC20-related workaround for netplan.io made by Simon re: dbus timeouts
    • Cherry-picked the change and will copy it over to the ubuntu-image (used by the coreXX builds) PPA
1 Like
13
  • Holiday Mon & Tues
  • Tested @jawn-smith’s Unicorn HAT packages, added tests and docs (unicornd looks too complex to package currently)
  • Packaged rpi-imager (LP: #1968921) after snap was unlisted
  • Investigating partition growth fix (LP: #1947311)
  • Worked on Build HAT packaging (unfinished)
  • Pi meetings
1 Like
14
  • @jawn-smith sponsored systemd upload for jammy, which contains patches for LP 1966381 and LP 1926860.
  • Started looking into errors preventing phasing of systemd version 245.4-4ubuntu3.16 to focal.
  • Gave a quick review on an MP for a systemd hwdb change for HP hardware
  • golang-github-valyala-gozstd vs libzstd (ppc64el) for proposed-migration
    • I believe this to be some kind of memory issue on the infrastructure. Testing in a POWER9 VM with similar resources (autopkgtest limited to 1 CPU and 1592 MiB mem), the issue is not seen. Ran a memory profile of the go tests, but nothing stood out. Skipping one test (TestNewWriterLevel) is one option to allow the autopkgtests to succeed.
  • ubuntu-release-upgrader bug (LP 1968607).
    • @bdmurray uploaded the fix to Jammy
    • Working on SRUs for focal and impish
1 Like
15
1 Like
16

Ubuntu 22.04

I fixed bugs before the release.

  • ubiquity 22.04.15:
    • Set GNOME keyboard layout during installation to prevent keyboard layout changes before typing username/password (LP: #1875062).
  • apport 2.20.11-0ubuntu82:
    • Catch zlib.error when decoding CoreDump from crash file (LP: #1947800)
    • Catch FileNotFoundError during process_report (LP: #1867204)
  • paramiko 2.9.3-0ubuntu1:
    • New upstream release (LP: #1968730).
    • Add support for SHA-2 variants of RSA key verification algorithms to support openssh >= 8.8p1-1 (Closes: #1007168, LP: #1961979)
1 Like