Previous status can be found here: Foundations Team Updates - Thursday 5 January 2023
ubuntu-image
- Addressed comments on PR to fix missing customization steps for prebuilt rootfs use cases
- Addressed comments on PR to fix relative file path handling
- Created PR to add compression support to rootfs artifacts
- Reviewed PR to use new snapd seed.manifest generation
- Investigated Feature request from IoT field team
- Changed snap recipe options to build on all architectures
- Meetings with CPC team to discuss building cloud images with ubuntu-image
- Fixed a bug related to adduser/useradd differences in debootstrap output
Miscellaneous
Distro
- python-werkzeug proposed-migration (merge flask)
- MIR: review webpack-fdo (Jammy), https://pad.lv/1973033
- MIR: investigate licensecheck build-deps: https://pad.lv/2002426
- sponsor systemd-hwe + SRUs: https://pad.lv/1999782, https://pad.lv/2002388, https://pad.lv/2002065
- systemd merge review + sponsor: https://code.launchpad.net/~enr0n/ubuntu/+source/systemd/+git/systemd/+merge/435202
Netplan
- LP bug grooming
- netplan status fixes + merge PR#290
- NetworkManager integration (unit-)testing & patch squashing + improved docs: https://github.com/slyon/NetworkManager/tree/netplan-nm-1.40
- initial inclusive naming review, PR#303
- debuging NetworkManager-Netplan integration issue, https://pad.lv/1998207
- review memory-leak fix review+merge PR#309
- netplan set regression fixes & cleanup PR#299
Misc
- candidate interviews & take-home tests
- Core-Dev endorsement for @enr0n \o/
flash-kernel
- added Lenovo X13s to database
nezha-boot0
- package has been retired in Lunar as it is replaced by u-boot-nezha 22.10
u-boot
- added RISC-V boot protocol to Focal. Now needs testing to complete SRU.
U-Boot for VisionFive2
- set up script for building from upstream
- tested installation on SPI flash
livecd-rootfs
- adjust installation for LicheeRV and Nezha D1 for updated u-boot-nezha. Now daily builds on Lunar are working again.
Installer for Lenovo X13s
- work started
Miscellaneous
- review written interviews
Includes partial backlog of past (short) week.
autopkgtest
- Investigate cause of high failure rate of s390x VMs
- Wrote a (stub) script to quickly spawn a number of VMs and check they’re reachable. @bdmurray refined the script.
- Issue identified as a failure to reach VMs on s390x
- During the week: periodic checks of the reachability status of bos02 s390x VMs, asking IS to intervene when not reachable
- Root cause not identified as far we can tell.
bootspeed measurements
- Investigate failure to delete LXD testbed VMs
- Cause identified as an LXD bug, filed https://github.com/lxc/lxd/issues/11261
- The bug is now fixed in git and (likely) in edge, but not yet in the candidate or stable channels
Auto upgrade testing
- We had a full partition on venonat. After freeing up some space the tests are passing again.
ISO testing
- The Lunar automated ISO tests are failing.
- Root cause appears to be https://bugs.launchpad.net/brz/+bug/2000806
- Had IRC discussion with the Breezy maintainer. The bug is now Fix Committed.
- The discussion also lead to the filing of https://bugs.launchpad.net/launchpad/+bug/2002568
- Prepared and uploaded breezy 3.3.1-1ubuntu1 to Lunar to bring fix the ISO tests. The package is currently in lunar-proposed/universe.
- I’ll take take of syncing 3.3.2 once uploaded to Debian
cloud-init
- Integration testing: implemented new cleanup logic for LXD containers and VMs
- New logic is merged
- Same will have to be done for the other platforms/clouds
Server Team
- Initial work for enabling integration testing of the team projects on the IBM cloud
- Cloud resources: general cleanup of stale cloud resources left by integration testing.
- Cloud resources: we need a net set of jobs to cleanup all the stale cloud resources automatically.
- First “cloud”: LXD contaienrs and VMs. Prepared new job in Jenkins doing this, PR is up.
- Bug triage:
- Cleared the triage backlog for the winter break: https://lists.ubuntu.com/archives/ubuntu-server/2023-January/009547.html
- Weekly triage rotation: https://lists.ubuntu.com/archives/ubuntu-server/2023-January/009550.html
- HA testing (fence-agents-base, resource-agents-base)
- Fixed issues with the jobs in Jenkins. Four test cases remain failing, will discuss those with the maintainer of those packages next week (currently on pto)
Ubuntu Server ISO downloads metric
- The anonymized proxy logs have been moved to another swift bucket (ps45 -> ps5)
- Asked IS for new credentials to access it
- Temporarily disabled the Jenkins job for this metric to lower the noise
Misc
- Testing of ROCKs / OCI images: ongoing discussion on handing maintenance over to the ROCKs maintainers.
- Did several essay reviews and an interview.
Partial week, catchup from break
systemd
- Investigated
systemd-fsckd
patches which are specific to Debian and Ubuntu (dropped from upstream years ago) to see if we can move the functionality outside of systemd - Uploaded systemd 252.4-1ubuntu1 (@slyon sponsored)
- Investigating bugs that could be SRU candidates
Other
- More systemd-hwe SRU related work
- Created my core-dev application: https://wiki.ubuntu.com/enr0n/CoreDeveloperApplication
- OOO next week, so I will continue the process after that
Misc
- Unusually high rate of hiring-related activities.
- Patched
guilt
for git 2.38 compatibility, sent to Debian (thanks @ginggs for the upload) - Reviewed and sponsored openssl 3.0.7-1ubuntu1 for @adrien
- Usual batch of PR reviews on apport.
+1 activity
python3-defaults transition
metalfinder
My guess is it’s a regression in mutagen rather than metalfinder. To help with
the regression, I triggered migration-reference tests that also failed, but a
deeper investigation is still warranted.
nipype (networkx)
The nipype autopkgtests fail, but the issue seems to be networkx being
incompatible with numpy 1.24. Filed LP: #2002660 as the package is under the
umbrella of the openstack team.
matplotlib
autopkgtest failure on s390x. Investigation ongoing.
gnat 12 transition
ahven FTBFS
I looked into the ahven FTBFS (LP: #2000876). I was able to reproduce it
reliably, and narrow it down to an issue in libgnat or libgcc_s in exception
handling.
Other
- for @jawn-smith: practiced timing my arrival in meetings.
subiquity
- Fixed LP: #2001906 [CI failure caused by github rate-limit] through PR 1527
- Opened PR 1531 to fix regression causing inability to create encrypted LVM in manual partitioning
- Opened PR 1532 to fix LP: #2000066 [crash when editing encrypted LVM that was created in guided storage screen]. Rewored to fix inconsistency between client and server.
- Added support for copying deb822 sources in dry-run mode: PR 1533
- Opened PR 1528 to enable Ubuntu Pro screens (will merge after the GA release)
- Focused on mirror testing (no PR opened yet)
distro
- Asked around to retrigger autopkgtests to unblock sysvinit. Still needs two autopkgtest runs on s390x packages.
Distro work
- Investigated proposed migration excuses related to libconfig-inifiles-perl
Netplan
- Worked on the MIRs required by the new netplan dependencies due to the
netplan status
command. - Addressed the issues found during code review on the inclusive naming PR (PR#303)
- Fixed a memory leak (PR#309)
- Fixed a bug related to vxlan configuration being added to the wrong place (PR#310)
- Investigated a bug related to a vxlan option that can’t be disabled with netplan (PR#311)
- Addressed the issues found during code review on the netplan+networkmanager integration PR (merged) (PR#1)
- Experimented a little with clang-format. We could use it to enforce code style eventually.
- Last tweaks for the openssl merge (LP: #1998942)
- Sent for review the spec for the alternative to crypto-policies
- Fixed coreutils autopkgtest failure so it can migrate (LP: #2002428)
- Above will also require a change in datefudge (or giving up on datefudge)
Distro
- tzdata:
- Uploaded tzdata 2022g-1 to Debian unstable and merged into lunar (2022g-1ubuntu1)
- A lot of bug triaging. This results in marking 15 bugs as duplicate of bug #1648556 (still open). There was also the tzdata package upgrade bug #1672111 (marked three bugs as duplicate) that was caused by a bug in the date command (from coretuils package) in Ubuntu 18.04 “xenial”.
- Reported performance problem #868395 upstream and triggered a lively discussion there. It’s a known, long-standing glibc bug.
- rdma-core:
- Uploaded 44.0-1 to unstable (and synced to lunar)
- Uploaded 44.0-2 to fix mispel FTBFS
- sbuild:
- Updated https://salsa.debian.org/debian/sbuild/-/merge_requests/36 to not introduce a regression
- Reworked unshare autopkgtest merge request
- d/t/get_default_release.py: Support Ubuntu
- apport: Upload 2.24.0-0ubuntu2 to fix test failures with Python 3.11
Apport
- feat: Move Java crash files into java/testsuite
- fix: Fix test failures with Python 3.11
- Make Apport use inclusive naming:
- fix: Ensure full coverage of test_get_file_package_uninstalled
- refactor: Rename TestSuiteUserInterface to UserInterfaceMock
- Skip tests if libraries are missing:
- test: Test _wait_for_process() helper method
- deb822 sources support for python-apt, almost full test suite coverage so far.
- grub 2022v1 signing getting ready
Happy New Year!
Archive
- Reviewed a new tool from Christian to identify removal candidates from the FTBFS reports. (Needs work, but a couple of old Ubuntu-only packages have gotten removed so far as a result.)
- Tracking
process-removals
(for packages removed from Debian unstable). Gotpostgresql-14
removed from lunar, and moved removal ofpython2.7
forward. - Reviewed general removal requests.
- Attempting to prune
python-cryptography
reverse-dependencies on i386 (livecd-rootfs
;curl
) to avoid having to bootstrap a stack of rust modules.
Release
- Checked in on status of 22.04.2 blockers
- Did some work to untangle transitions in lunar-proposed. Qt transition has landed, and packages entangling Qt with
python3-defaults
have been copied back. Rolled backprotobuf
to not entangle; now trying to getllvm-toolchain-15
releasable to get theqt6-base
transition through, then all packages should be ok to copy back to lunar-proposed again for the bigpython3-defaults
transition. - Code reviews for the proposed official Xubuntu Minimal image
SRU
- Released the last round of grub SRUs for focal, unblocking an arm64 boot blocker
- Reviewed the latest round of grub SRUs for jammy/kinetic (and accepted grub2-unsigned), for CVEs.
General
- Did an interview for a toolchain squad candidate
- Helped Legal generate some reports used for copyright registration of Ubuntu
- Discussions around SecureBoot signing with the latest keys for the new SBAT level in the latest security updates
- R(e-r)eview of a couple of specs for @bdrung
- Finished lsb merge; ready for review & sponsorship
by some poor masochist(LP: #2002318) - Reviewed and sponsored @xypron’s changes for RISC-V EFI for focal (LP: #1998513)
- Investigating Xilinx bootgen changes (LP: #1998728, LP: #2000773, LP: #2000774)
- Investigated libcamera status with new Pi Camera Module 3 (needs work) under Ubuntu
- Pi meetings
- Interviews & tests
autopkgtest
- Merged an MP from @enron adding dpdk to big_packages for arm64.
- While testing the
--archicture i386
with autopkgtest I encountered an issue trying to test gzip which was already reported in a bug. - Worked with @paride and IS to resolve an issue with a bad s390x host in bos02. Bad in that it was creating instances that were unreachable.
- Reported a bug about inconsistent logging in the command ssh-setup/nova. Sometimes the error and debug_failure functions are invoked and sometimes they are not - or at least we don’t see them in journalctl on the could-worker.
- Submitted an RT regarding our inability to create new s390x instances in bos01. Pinged IS regarding the RT I submitted. Dug a bit further and noticed that the instances are initially being sent to swirlix.
- Asked IS to bounce services again in bos02 as s390x instances created on s0lp4 weren’t reachable.
- Submitted an MP adding a tool to create test instances in a region for an architecture to aid in debugging.
- Investigation into how long it takes for
ssh-setup/nova
to run and experimentation with the wait period to create new instances.
Ubuntu
- Reviews of candidate essays.
- Conducted an interview x 2.
- SRU team shift.
-
dpkg
- Working on duplicated LTO flags in dpkg-buildflags. https://pad.lv/2002201
- Fix regression caused by merge of 1.21.17 https://pad.lv/2002582
-
New point release for golang-1.18 & golang-1.19
- golang -1.18 1.18.10-1 has uploaded to Debian and synced to Ubuntu.
- golang-1.19 has filed unblock request pre-approval in Debian.
-
golang-1.20 rebuild
Still triage failures
- https://pad.lv/2002625 go list: lstat …/…/share/go-1.20/src: no such file or directory
- Upload golang-github-crowdsecurity-machineid 1.0.3 to fix ftbfs
- Upload synthing 1.19.2~ds1-1 to get rid of golang-github-marten-seemann-qtls-go1-18
- Request removal of golang-github-wellington-go-libsass golang-github-bep-go-tocss in Debian, which are FTBFS with Go1.20 and are obsoleted libraries.
-
Go packages proposed migration
To get some packages updated so they are not FTBFS
- Upload golang-github-containers-storage 1.43.0+ds1-7 to build with old btrfs-progs, to decouple the migration with btrfs-progs.
- Upload golang-google-grpc 1.33.3-2 to fix autopkgtest in Ubuntu
-
openjdk:
- prepared a list of jtreg test failures for analysis
- added/updated merge requests on salsa for jtreg environment fixes (obvious problems with tests)
- added draft implementation in ppa for failing build on test failure
-
ca-certificates-java:
- cleaned up package’s bugs
- validated updated package for https://bugs.launchpad.net/ubuntu/+source/ca-certificates-java/+bug/1998065
-
asmtools:
- new package for https://github.com/openjdk/asmtools
- submitted ITP and package to debian, currently in new queue