Tested, finished and merged MP with script for cleaning up janky results in swift
Deployment of new charms
MP review for Brian and Paride
Talking with IS, trying to figure out an issue with users viewing the containers holding our test results - there’s a limit to the amount one can see, and there is seemingly no way to work around it!
quick bugfix for download-all-results
MP modifying metrics script to include the percentage of worker units which are active - this is then used to set up a new alert in grafana to help ping us when there are issues with a specific datacentre/arch
desktop testing
Retriggering the smoke default test for oracular, as it was tempfailing
Meeting with Cert Team to sync up regarding the hardware-installer-testing - sharing pain points and discussing next steps
Testing all of my various test cases - progress here
britney
Investigating how many i386 tests we run which are actually unnecessary
I was on vacation from June 27 to July 4, so this has been a particularly short pulse. The status also includes work done in the last week of June, since I was off the last two Thursdays.
dotnet
Created stage 2 .NET 8 package for ppc64el Oracular Oriole (PPA)
The package has been uploaded to Oracular and is now available in the archive.
Attended the weekly .NET partners sync meeting.
Updated .NET snap spec to include the upgrade story for current users of the dotnet-sdk snap.
Reviewed and responded to comments on the .NET snap spec (FO151).
Investigated how the .NET SDK resolver uses the dotnet symlink realpath and what alternatives could be implemented in the code to tend to the snap use case where the dotnet executable on PATH does not link directly to the dotnet CLI executable, but to the snap executable (relates to canonical/dotnet-snap#13).
Released .NET 6 and .NET 8 July updates for the .NET Runtime content snaps (dotnet-runtime-60 and dotnet-runtime-80), the .NET SDK content snaps (dotnet-sdk-60 and dotnet-sdk-80), and the legacy .NET SDK snap (dotnet-sdk).
Also worked on creating .mount unit files for the content snaps. These will be shipped along with .NET binaries so that the .NET installer can easily mount their locations (this is still an investigation in progress).
Worked on bits for the .NET Installer:
Created mount and unmount commands to mount/unmount content snap bind-mounts to the dotnet installation directory.
Created the .NET content snaps repository, which aggregates all the runtime and SDK content snap definitions in one location. Also set up GitHub Actions workflows to automatically publish the snaps to the store upon push.
Created the .NET manifest repository and content snap listing in the store.
fixed two small issues with copyright and changelog.gz symlinks
added basic crypto-config integration with no profile so far
the default configuration now reads dropins from /etc/ssl/openssl.conf.d but I now notice I typo’ed and wanted to use .cnf.d instead to match openssl.cnf
Investigated support for “include” directives in gnutls, proposed a patch upstream in order to kickstart the discussion (I don’t want to introduce something and have upstream introduce something different)
still wondering what “ is not a candidate” means in excuses; next week I’ll assume nobody knows (wouldn’t surprise me)
attempts to copy a long and moderately complex specification from google docs to discourse; the pain is real
couldn’t find a converter to (discourse) markdown that handled both line-feeds in tables and strike-through text
copying HTML is actually OK-ish but CSS classes need to be inlined into nodes
asked about enabling an auto-toc plugin and Aaron Prisk has forwarded the request (there’s an official theme component for that with plenty of features, e.g. sticky nav)
Completed tck-17 and -21 runs in lxd containers on amd64 (ff, jj, nn). Found a test failure only in lxd, not on VMs, due to a behaviour difference in fsetxattr() – needs more investigation. Patched the build-tck-lxd and runtck scripts with minor fixes
drafted a curtin MP for installing dracut when NVMe/TCP is used. Needs more work since dracut get automatically removed at the end of the installation.
created small dracut module to feed network configuration, extracted from the NBFT, to systemd-networkd
rust-reqwest also needed additional help to handle proxies: upstream PR (thanks @schopin for the help during investigation) Patch is still waiting for a Debian sponsor.
apport-retracer
Drafted specification FO191 covering automated testing of apport-retracer.
MP: #469060, MP: #469061 working on getting noble builds running. Curtin itself seems fine but CI is unhappy. First part merged, sorting out ppc64el test runner issue.
MP: #468543 - minor doc tweak to clarify github vs lp for curtin location
subiquity
PR: #2024 - merge autoinstall feature for noble related to disk matcher support
PR: #2022 - Stable branch merge to Noble for the changes to address LP: #2063331
PR: #2026 - Introduce debug mode in Subiquity for wrapping ubuntu-drivers with umockdev wrapper script to test driver installs in a virtual machine
Analysis of Subiquity’s usage of cloud-init’s python module and what it would take to have subiquity only rely on cloud-init outside of the snap. (To avoid surprising bugs like LP: #2062988)
Daily bug triage
Distro
Begun investigating the necessary changes required to SRU a fix for LP: #1974483 to Jammy. I have opened a merge-proposal against ubuntu-seeds as a start.
Pair programming session with @vpa19771, trying to debug an issue with piuparts in oracular that, for whatever reason, decided to resolve itself after firing up the debugger