essentially a bit less than half the time available to work on +1
maintenance.
# 2025-07-14
## gsasl + dovecot
gsasl autopkgtests are failing with dovecot:
[doveconf: Fatal: Error in configuration file
/tmp/tmp.DXPuLQ1W9A/d/dovecot.conf line 1: Unknown setting:
dovecot_config_version]
Fixed in https://launchpad.net/ubuntu/+source/gsasl/2.2.2-2ubuntu1 ,
just need the right trigger, and infrastructure cooperation (`nova [W]
nova quota exceeded (attempt #1)`)
Triggers sorted it for amd64. Doing the same for all other
architectures now, in the hopes of resetting the baseline to green.
## apparmor and ssh-keygen profile
All tests with this signature:
Saving key "/tmp/autopkgtest.xr2SAo/autopkgtest_tmp/id_rsa" failed:
Permission denied`
Should be retried with trigger apparmor/4.1.1-0ubuntu4 because that's
where it is fixed
(https://launchpad.net/ubuntu/+source/apparmor/4.1.1-0ubuntu4):
$ ./retry-autopkgtest-regressions --log-regex "Saving key.*failed:
Permission denied" --add-trigger apparmor/4.1.1-0ubuntu4
## livecd-rootfs build errors
Weird.
dh_compress
sh: 1: cannot create /dev/null: Permission denied
...
/usr/bin/dpkg-deb: line 25: /dev/null: Permission denied
Checked a local questing-proposed vm, builds fine. PPA reproduces the
build error from the archive. Checked apparmor changelog, and also
updated it, but no errors in the local build.
# 2025-07-15
## openssh still failing due to apparmor?
93s Key is a ssh-rsa key
93s Wrote key to '/etc/dropbear/dropbear_rsa_host_key'
93s ssh-keygen: /tmp/dropbear_rsa_host_key-c0Fl11sf45.pub: Permission denied
But I triggered a run with the fixed apparmor package, so what is
going on? Investigating…
UPDATE: a rm_conffile is missing in apparmor 4.1.1-0ubuntu4: it just
stopped shipping the ssh-keygen profile, it didn't deactivate it. So
just an upgrade won't fix the problem. But since yesterday, apparmor
migrated, and new test VMs today are bootstrapping with that
installed, so the bug is no longer today, and test runs triggered
today are passing.
## livecd-rootfs FTBFS
Seems like it's rust-coreutils at fault:
https://bugs.launchpad.net/ubuntu/+source/rust-coreutils/+bug/2116973
root@q-mknod:~# rm null
root@q-mknod:~# mknod --mode 0666 null c 1 3
root@q-mknod:~# l null
crw-r--r-- 1 root root 1, 3 Jul 15 18:36 null
I filed a bug upstream: https://github.com/uutils/coreutils/issues/8342
I went through the NBS report (I'm going through one by one, instead
of copy&pasting the command from the NBS report, and checking the
rdepencencies by hand, at least initially):
kdevelop62-libs, libcamera0.4, libdb++-dev, libmiroil3t64, libmiral5t64
# 2025-07-16
## New queue
### mutter
Got a ping about it at
https://matrix.to/#/!HIqUfDuodVisBWdrTr:ubuntu.com/$YZhsZoWlMFyhdU8tN8-GtKCQGXJqAgbErc3cPypWeHQ?via=ubuntu.com&via=matrix.org&via=matrix.debian.social
Accepted after overriding main → universe for mutter-dev-bin.
## NBS
Still doing a few manually, to slowly get more confidence.
Done: libmirserver58t64, libmircookie-dev, libmircookie2t64,
libnutscan2, libokular6core3, libqmobipocket2, libqmobipocket6-2,
mir-platform-graphics-eglstream-kms20,
mir-platform-graphics-gbm-kms20, mir-platform-graphics-wayland20,
mir-platform-graphics-x20, mir-platform-input-evdev8,
mir-platform-rendering-egl-generic20, mir-doc
## Excuses
### apparmor fallout
Another round of retriggers due to the apparmor fallout. Just
upgrading the apparmor package isn't enough, because the "fix" was to
remove the ssh-keygen profile file from the package. Upgrading to the
new package won't remove that file from the system, because it's a
config file. Version 4.1.1-0ubuntu5 was uploaded with an rm_conffile
addendum. While that version is not yet available, fresh installs of
0ubuntu4 should be sufficient, and by now the VMs used by the
autopkgtest infrastrucure should have it.
So another round of:
$ ./retry-autopkgtest-regressions --log-regex "Saving key.*failed:
Permission denied"
## open-iscsi autopkgtest error
197s + apt-get install -qy --only-upgrade glibc-doc=2.41-6ubuntu2
glibc-source=2.41-6ubuntu2 libc-bin=2.41-6ubuntu2
libc-dev-bin=2.41-6ubuntu2 libc-devtools=2.41-6ubuntu2
libc6-amd64=2.41-6ubuntu2 libc6-dbg=2.41-6ubuntu2
libc6-dev-amd64=2.41-6ubuntu2 libc6-dev-i386=2.41-6ubuntu2
libc6-dev-x32=2.41-6ubuntu2 libc6-dev=2.41-6ubuntu2
libc6-i386=2.41-6ubuntu2 libc6-x32=2.41-6ubuntu2 libc6=2.41-6ubuntu2
locales-all=2.41-6ubuntu2 locales=2.41-6ubuntu2 nscd=2.41-6ubuntu2
197s Reading package lists...
197s Building dependency tree...
197s Reading state information...
197s Package libc6-amd64 is not available, but is referred to by
another package.
197s This may mean that the package is missing, has been obsoleted, or
197s is only available from another source
197s However the following packages replace it:
197s libc6
I don't understand it. The libc6-amd64 binary package does seem to
exist, but at the same time doesn't? I asked in matrix
(https://matrix.to/#/!QMtJBibTYYOCvXJEdv:ubuntu.com/$wLNpKG0b0MY6laEQUQzPZZdcfeAfvLm7yCrHiEEvtBY?via=ubuntu.com&via=matrix.org&via=matrix.debian.social).
# 2025-07-17
Today is my SRU shift, and there are many SRUs pending. This is made
worse by the fact we have two SRU team members on PTO, meaning the
only SRU days this week are today, and yesterday.
# 2025-07-18
## SRU
Finished a spillover from my SRU shift yesterday (snapd for noble and jammy).
## open-iscsi autopkgtest failure
I filed this bug
https://bugs.launchpad.net/ubuntu/+source/open-iscsi/+bug/2117102 the
day before my SRU shift, and today I want to pick it up again.
Got to chat about it in MM, have some ideas, was setting up a local
reproducer (had to workaround some things like tmpfs being too small
for this test), and now I have to jump into patch pilot, so no real
progress here this time.
--
ubuntu-devel mailing list
ubuntu-devel@lists.ubuntu.com
Modify settings or unsubscribe at: https://lists.ubuntu.com/mailman/listinfo/ubuntu-devel
