+1 maintenance _day_ report
paride at ubuntu.com
Wed Jun 22 13:32:33 UTC 2022
As part of the newly formed Ubuntu QA team  I'm doing a shorter 1-day
+1 maintenance rotation which, other than helping with the usual +1
duties, is a practical way for me to monitor the state of the
autopkgtest infrastructure. Here is the summary for this week.
- plink1.9, stuck in kinetic-proposed for >130 days due to an
autopkgtest regression. Failure identified as an LTO issue. Package
added to lto-disabled-list (LP: #1979247), followed by a no-change
upload. It migrated.
- firejail, stuck in proposed due to autopkgtest regression on armhf.
Issue identified as due to new tests requiring (at least) a privileged
container to success, while the Ubuntu armhf testbed system is an
unprivileged LXD container. I uploaded a workaround (restricting the
failing test to isolation-machine, see LP: #1979358), and also filed
Debian bug on this . After a fruitful discussion with the Debian
maintainer I merged a new package version from Debian, this one
requiring the workaround (tested by triggering autopkgtests from a PPA;
package currently in -proposed). I also suggested a possible way to drop
the only delta the package currently has.
- I polished some local changes I had to allow autopkgtest-virt-lxd to
run tests in LXD VMs, and submitted this upstream . I think
testing/feedback from Ubuntu developers is needed there. If that gets
merged we could consider running the armhf tests in LXD VMs, avoiding
issues like the firejail one I mentioned above.
- I submitted another MR to autopkgtest upstream to update the
documentation to make it explicit that the isolation-container
restriction does not imply that the container is privileged.
- I updated the ProposedMigration wiki page  to suggest that strange
FTBFS/crashes in newer Ubuntu releases not seen in Debian may be due to
LTO, and that it's worth checking.
More information about the ubuntu-devel