+1 Maintenance Report

Thu May 30 22:52:05 UTC 2024

On Wed, May 29, 2024 at 07:04:11PM +0000, Graham Inggs wrote:
> I was on +1 maintenance from 2024-05-23 until 2024-25-29.  Below are
> the things I worked on.
> 
> 
> The autopkgtest queues were completely empty following the Mardid
> sprint.  I noticed the update_excuses report [1] still showed many
> running tests in various states.  I was able to re-queue these using
> retry-autopkgtest-regressions from ubuntu-archive-tools [2] (and
> piping the output to vipe, etc. as described in the help text).
> 
> Retried 5222 tests in state "Test in progress":
> ./retry-autopkgtest-regressions --state=RUNNING
> 
> Retried 374 tests in state "Reference test in progress, but real test
> failed already":
> ./retry-autopkgtest-regressions --state=RUNNING-REFERENCE --no-proposed
> 
> Retried 359 tests in state "Test in progress (will not be considered a
> regression)":
> ./retry-autopkgtest-regressions --state=RUNNING-ALWAYSFAIL
> 
> I encountered some 503 errors while queueing these up, so it took a
> few tries to get them all queued.
> 
> 
> r-base / rmatrix
> ================
> sync'd r-cran-tmb
> https://launchpad.net/ubuntu/+source/r-cran-tmb/1.9.11-2
> 
> r-cran-glmmtmb needed a no-change rebuild against r-cran-tmb and rmatrix
> https://launchpad.net/ubuntu/+source/r-cran-glmmtmb/1.1.8+dfsg-3ubuntu2
> 
> r-cran-bbmle's autopkgtest regressed on ppc64el only, but it had
> already regressed in Debian testing, so I filed Debian bug #1071667
> and added a hint
> https://git.launchpad.net/~ubuntu-release/britney/+git/hints-ubuntu/commit/?id=5bff2fb75385c247d0a9953c4830e8235d21ec1a
> 
> r-cran-mertools' autopkgtest was failing on ppc64el only with:
> ! R session crashed with exit code -9
> but passed in Debian.  The test succeeded for me on a ppc64el cloud
> instance, so I tried adding it to big_packages and it passed
> https://git.launchpad.net/~ubuntu-release/autopkgtest-cloud/+git/autopkgtest-package-configs/commit/?id=bc49922e209d3c333469177cf81f6b461245788d
> 
> This was enough for r-base and rmatrix to migrate.
> 
> 
> octave
> ======
> octave-statistics FTBFS with a dwz error, uploaded a workaround
> https://launchpad.net/ubuntu/+source/octave-statistics/1.6.6-2ubuntu1
> 
> nlopt's autopkgtest was failing when nlopt was built with
> -D_FORTIFY_SOURCE=3, uploaded a workaround
> https://launchpad.net/ubuntu/+source/nlopt/2.7.1-5ubuntu1
> 
> octave-fits FTBFS with octave 9 and was removed from Debian, filed LP:
> #2067300 for its removal from Oracular
> https://bugs.launchpad.net/ubuntu/+source/octave-fits/+bug/2067300
> 
> 
> astropy
> =======
> astroquery FTBFS and was removed from Debian testing, filed LP:
> #2067093 for its removal from Oracular
> https://bugs.launchpad.net/ubuntu/+source/astropy/+bug/2067093
> 
> 
> bowtie2
> =======
> bowtie2's autopkgtest failed on s390x because it is not installable
> there.  I triggered a migration-reference/0 test which failed, and
> bowtie2 migrated.
> 
> 
> black
> =====
> black's autopkgtest failed on i386 because the package went from
> arch:all to arch:any
> and is not built on i386, added a hint
> https://git.launchpad.net/~ubuntu-release/britney/+git/hints-ubuntu/commit/?id=28e9f6676dccc3f765c446241556561dd82c934c
> 
> html2text
> =========
> html2text's autopkgtest only passed on amd64, but it had already
> regressed in Debian testing and was reported in #1069872, added a hint
> https://git.launchpad.net/~ubuntu-release/britney/+git/hints-ubuntu/commit/?id=5fd1ec260fc922abc14ca89391533685618645fd
> 
> 
> feature-check vs confget
> ========================
> confget's autopkgtest regressed with the new version of feature-check,
> but it had already regressed in Debian testing, so I filed Debian bug
> #1071683 and added a hint
> https://git.launchpad.net/~ubuntu-release/britney/+git/hints-ubuntu/commit/?id=730a71a2e887c5f21e3161b2bdbc9da8b229aaeb
> 
> feature-check's own autopkgtest failed on i386 because the package
> went from arch:all to arch:any and is not built on i386, added a hint
> https://git.launchpad.net/~ubuntu-release/britney/+git/hints-ubuntu/commit/?id=5b0c5de382794d8b698fd1e3d90042eb36f50cd1
> 
> feature-check did not migrate until vorlon uploaded a no-change rebuild
> https://launchpad.net/ubuntu/+source/feature-check/2.1.0-2build1
> 
> I'd love to know what / where / how he spotted the problem.
> 
> 
> python-libarchive-c
> ===================
> python-libarchive-c's autopkgtest failed on i386 due to missing python
> extensions, added a hint
> https://git.launchpad.net/~ubuntu-release/britney/+git/hints-ubuntu/commit/?id=219f0c672e6f4df81f8625607f44add741b3c87f
> 
> 
> dipy
> ====
> dipy's autopkgtest regressed on armhf, but also regressed in Debian
> testing, so I re-opened #1071448 and added a hint
> https://git.launchpad.net/~ubuntu-release/britney/+git/hints-ubuntu/commit/?id=f3e3e4df84c1641c55570b0797b9667ac4d82fe9
> 
> 
> various syncs
> =============
> I looked at the Merge-o-Matic report for packages in universe [3], and
> sync'd several packages from Debian where our delta could be dropped:
> swig, indexed-gzip, python-djvulibre, openstructure, astrometry.net,
> virulencefinder, scanssh, afflib, netcdf, ipyparallel, mercurial, slrn
> and libflorist
> 
> mercurial still FTBFS on the buildds with some test failures, so I followed up
> https://launchpad.net/ubuntu/+source/mercurial/6.7.3-1ubuntu1
> 
> big_packages
> ============
> Lastly, I looked for possible candidates for big_packages by running:
> ./retry-autopkgtest-regressions --log-regex Killed
> 
> which returned the following:
> python-anndata, pythran, openmm, pandas, python-redis, netplan.io,
> openvswitch and samba
> 
> (I had tried this earlier in my shift, but ran into a 404 error due to
> a missing autopkgtest log, which I retried.)

The 404 you encountered is a consequence of us only copying forward
neutral and passing autopkgtest results when opening the archive for a
development release of Ubuntu. However, britney (which generates
update_excuses) is not aware of this and produces a non-existent link
for the log file. Regardless, retry-autopkgtest-regressions should not
crash just because one log file is not available and I've fixed that
now[1].

[1]
https://git.launchpad.net/ubuntu-archive-tools/commit/?id=3fc993e971fcbcb9ebb5729a30e56eed72514b5f

Cheers,
--
Brian Murray