The previous status is here: Foundations Team Updates - Thursday 16 March 2023
ubuntu-image
- Lots of follow up and discussion for preseeding classic images
- Our approach seems fine, but snapd 2.59 has a bug that is preventing this from landing: https://bugs.launchpad.net/snapd/+bug/2012567
- Lots of updates to the update bootloader PR
- Opened PR to reduce tech debt by removing any references to the deprecated io/ioutil package in Go
- Had meeting with field team about new features. One was trivial to implement, so I opened a PR to expose the sysfs overlay option to snap prepare-image
- Lots of work on getting the preinstalled Raspi desktop images in a good state. Almost there
Misc
- proposed migration: e2fsprogs
- uploaded a new version of fsarchiver with a patch from upstream, which allowed e2fsprogs to migrate
- Take home test reviews
- Interviews
1 Like
- Finished no-change rebuilds for the python3.11-only transition
- Sponsored dotnet6 and dotnet7 March 2023 updates for @mirespace
- Sync’d pymatgen and added it to big_packages
- Sponsored syncs of golang-golang-x-tools, packer and git-sizer for @zhsj
- Sponsored dotnet6 and dotnet7 uploads, as well SRUs for kinetic and jammy, for @dviererbe
- Candidate interviews
2 Likes
Embedded World
- We were present at the RISC-V booth and I gave a talk on the status of Ubuntu on RISC-V.
gnome-shell
- Gnome shell stopped to work on RISC-V due to an upstream change.
- Analysis showed that the .dynamic section of RISC-V and MIPS is read-only and therefore not relocated in the glibc routines running at _start. https://gitlab.gnome.org/GNOME/gnome-shell/-/issues/6528 @duflu now has provided a patch which fixes the problem.
gdb
- On Qualcomm 8cx Gen 3 SC8280XP gdb fails read pointer authorizations registers. As this is also reproducible in a simple C program it seems to be a issue with the kernel. LP #2012509
apport-retrace
- gdb started by apport-retrace failed to support the layout command. Analysis showed that environment variable $TERM was not passed. @bdrung provided a patch.
box64
- I analyzed the usability of box64 for emulation on RISC-V
- Emulation of binaries requires adding amd64 as additional architecture and installing dependent amd64 libraries.
- For complex C programs like GLPK and the U-Boot sandbox the emulation was successful but slow (factor 80 compared to a native RISC-V program).
- For some Gnome amd64 libraries the postinst phase of package installation failed.
- For C++ libraries some symbols could not be resolved when trying to run the applications.
- I was neither able to run a Qt5 binary (Kcalc) nor a Gtk binary (Gnome-calculator).
- box64 is incompatible with qemu-user-static: only one of them can be the default emulator to execute amd64 binaries
1 Like
A week for snaps.
-
Go snap
- Transfer the publisher to Canonical.
- Refine the description, asciinema video.
- Enable building Go1.19 and Go1.20 on riscv64. But Go1.20 fails to build. Will investigate on next week.
-
Docker snap
- Merged PR
- WIP PR
I would need some help from apparmor&kernel experts for python installation issues on docker snap.
1 Like
autopkgtest
- Added linux-starfive to never_run for Lunar because we don’t run riscv64 tests
- Added nvidia-cuda-toolkit to big_packages to see if it would help with ENOSPACE, while it did help the test still failed.
- Updated README.md in autopkgtest-package-configs with information about big_packages and disk space.
- Started autopkgtests for jinja2 on an autopkgtest-cloud-worker unit as they couldn’t be run otherwise due to there not being previous results. This was brought up by @ginggs.
- Gave @paride commit access to the ~ubuntu-release autopkgtest repository.
- Pinged IS regarding restarting neutron on s0lp4 (bos02 s390x host) as the s390x queues are slow. The underlying issue is tracked in an RT.
Error Tracker
- Tested the staging version of the Error Tracker with cassandra 3.11.
ISO Tracker
- Updated two test cases on the ISO Tracker server regarding RAID server installs for @dbungert.
Ubuntu
- Updated the Lunar Release schedule to have a link to the BetaFreeze wiki page and also the template tool which is used to create the Discourse post.
- Submitted an MP for retry-autopkgtest-regressions so that it will not Traceback when processing stable releases which have private items in the queue or running.
- ISO testing for the 20.04.6 point release.
- Update the meta-release files for the 20.04.6 point release.
Misc
- Take home test reviews
- Interview(s)
1 Like
Distro
- +1 maintenance week
- Migrated pycodestyle
- Fixed update-notifier autopkgtests
- Working on icdiff autopkgtests so python-flake8 can be migrated. Worked with upstream to fix the problem and submitted a PR so we can drop all the patches on the next release
Netplan
1 Like
Distro
- MIR: log4cplus ACK, https://pad.lv/2003549
- MIR: drop new B-Ds of ndctl, https://launchpad.net/ubuntu/+source/ndctl/76-1ubuntu1
- MIR: review libtraceevent (NACK), https://pad.lv/2009715
- MIR: libtracefs NACK, https://pad.lv/2008799
- inetutils-telnet review:
- https://code.launchpad.net/~dviererbe/ubuntu/+source/inetutils/+git/inetutils/+merge/439257
- sponsored inetutils[-telnet] for @dviererbe, https://pad.lv/2009814
- systemd sponsoring (Lunar, Kinetic, Jammy):
- MIR PRs:
- review & merge, https://github.com/canonical/ubuntu-mir/pull/10
- review & merge https://github.com/canonical/ubuntu-mir/pull/13
- create https://github.com/canonical/ubuntu-mir/pull/14
Netplan
- NetworkManager integration bugs:
- temp OpenVPN connection profiles, https://bugs.launchpad.net/ubuntu/+source/network-manager/+bug/1998207
- NM snap crash with snapd integration, https://bugs.launchpad.net/ubuntu/+source/netplan.io/+bug/1959570
- NM+Netplan connection migration, https://bugs.launchpad.net/ubuntu/+source/network-manager/+bug/2010561
- config spec discussions & approval
- investigate & fix DebCI vs systemd v253, https://bugs.debian.org/1033265
- NetworkManager snap review
- review spread test integration, https://github.com/canonical/netplan/pull/330 (cc @mvo)
- prepare Ubuntu Core/snapd PATH fixes, https://github.com/canonical/netplan/pull/336
Misc
- take-home-test reviews
- candidate interview
- M&G Interview template
- Roadmap grooming
2 Likes
subiquity / curtin / probert
- merged PR 1586 to fix a collision between a method and an attribute (merged to ubuntu/lunar)
- fixed LP: #2012416 and duplicate LP bugs through PR 1601 (merged to ubuntu/lunar)
- ongoing investigation of failed installs caused by a crash in pyudev (called from probert). The theory is that the crash would be triggered by non UTF-8 partition labels ; and more specifically NTFS partitions on french MS Windows installs. To be confirmed. I don’t have a reproducer and the lack of a system with such configuration makes it difficult to confirm. Manually changing the NTFS label to did not trigger the crash.
- ongoing investigation of failed installs caused by the use of the ventoy software solution, which interferes with the linux device mapper. Can not reproduce yet but I got bug reports from at least 3 different people already. Will do more testing in due time. I suspect that building a ventoy image from Linux and from Windows ends up being different.
- code reviews
misc
- spec reviews
1 Like
systemd
- Working on spec for ManagedOOMPreference integration
- Follow-on uploads for kinetic, jammy, and lunar to complete fix of LP: #2002445
- Working with Debian to get libsystemd.a shipped in libsystemd-dev (LP: #2012437)
software-properties
- Working on adding deb822 support to add-apt-repository
- Most work so far has been on removing the “extended” sourceslist API that wraps python-apt
1 Like
- Tested kernels for an HDMI audio bug on Jammy (LP: #2009136)
- Uploaded
flash-kernel
fix preventing flashing of not-yet-completely-installed kernels (LP: #2007827) - Broke
flash-kernel
with the aforementioned fix (LP: #2011925); uploaded tentative fix to PPA for testing - More testing of candidate core22 build of firefox (hardware compositing now working; testing full Wayland compatibility)
- Tested lunar pre-installed desktop images for @jawn-smith
- Tested 4K 60Hz support on pre-installed desktop images (works, but requires a bit of hoop jumping; requires further documentation)
- Tested new Pi global shutter camera with libcamera stack (works out of the box on lunar)
- Pi meetings
1 Like
Most of the other tasks are currently being blocked by external factors.
Rust
- Currently getting the proposal to merge
cargo
andrustc
source packages reviewed
Ubuntu ROCK
- Working on benchmarking the Chiselled images for Rust application deployment
1 Like
LP #2008789: [MIR] inteutils-telnet
- Fix Commited for LP #2009814: Inetutils is failing almost all autopkgtests. MIR is still in review.
dotnet
- dotnet6:
- added runtime IDs for kinetic to lunar release
- LP #2011807: [SRU] backport dotnet 6.0.115 to kinetic and jammy
- dotnet7:
- added runtime IDs for kinetic to lunar release
- LP #2011809: [SRU] backport dotnet 7.0.104 to kinetic
- started working on adding more deb8 tests
- looking into the dotnet snaps
Other
- 20.04.6 image testing; discovered bugs:
- LP #2012350: Screen Reader stopped working after launching Live Session
- LP #2012346: Live Session Testcase instructs to not select the Live Session
- LP #2011789: Battery state displayed as “Not charging” although laptop is plugged in and hardware indicator light is on.
- LP #1845362: gnome-font-viewer crashed with signal 5 in _XEventsQueued()
- minor reviews of ubuntu-mir documentation PRs
Special Thanks again to @ginggs, @slyon, @mirespace!
1 Like
- subiquity
- LP: #2006985 - add new livecd-rootfs hacks to adjust journald flags for compatibility with core22. At least we can turn zstd back on. Merged, uploaded, confirmed fix on daily build.
- PR: 1606 - workaround a hang when using autoinstall with desktop.
- PR: 1607 - merge to lunar
- PR: 1608 - make sure both series linters run, both to work around current flake8/pycodestyle issues but also it seems like an improvement
- LP: #2008952 - investigate systemd ordering a little bit. Even outside of the autoinstall case the subiquity process is starting later than expected for u-d-i - 40+ seconds later for at least oen user.
- mini-iso-tools
- start handling MIR team feedback from LP: #2008742
- reviews on cdimage changes, debugging of livecd-rootfs builds
- probert
- PR: 131 - fix logging of filesystem resize tool error output. This is also a template for further logging improvements I want to do generally, with the goal that stdout/stderr/returncode of such tools are logged consistently in and in a consistent location.
- sponsor openjdk-lts upload for @vpa1977. Wrap my head around git-ubuntu rich history as part of that.
- proposed migration libzstd - failure of rpmlint needed python-zstandard trigger added, i386 handled through hint per faq.
- 23.10 planning
- greenhouse
1 Like
Distro
- devscripts: Drop debpkg to make devscripts architecture all. Instead of debpkg please use sudo for debi (or, highly dangerous, make dpkg setuid root). Uploaded 2.23.3ubuntu2 to lunar.
- python-tz:
- Dynamically determine list of available and common timezones (LP: #207604)
- Determine IANA (nee Olson) database version dynamically
- Add autopkgtests to run unittest and own regression tests
- Upload 2022.7.1-2 to unstable
- Prepared and uploaded 2022.2.1-1ubuntu0.22.10.0 for kinetic
- Prepared and uploaded 2022.1-1ubuntu0.22.04.1 for jammy
- tzdata: 2023a was released (LP: #2012599). I uploaded the new version to unstable (2023a-1), experimental (2023a-1exp1), and lunar (2023a-1exp1ubuntu1). Now I am preparing the update for the stable releases.
Apport
1 Like
Short week due to minor health issues.
- Another fun glibc/GCC UB investigation
- autopkgtest investigations following last week’s glibc 2.37-0ubuntu2 updates
- Fixed a Rust 1.67-related FTBFS
- Take-home test grading
- Roadmap planning
1 Like
Distro
- SRU for anacron disabling itself upon upgrade (waiting for review) : https://bugs.launchpad.net/ubuntu/+source/anacron/+bug/2006589
- strace 6.2 (to be applied on top of change for strace 6.1); both are still waiting for review : https://bugs.launchpad.net/ubuntu/+source/strace/+bug/2012258
- some bug triage
- some preparations for openssl 3.1 (not for lunar)
1 Like