Your message dated Tue, 8 Jul 2025 10:13:11 +0100
with message-id <[email protected]>
and subject line Re: Bug#1093969: mutter: autopkgtest are flaky
has caused the Debian Bug report #1093969,
regarding mutter: autopkgtest are flaky
to be marked as done.
This means that you claim that the problem has been dealt with.
If this is not the case it is now your responsibility to reopen the
Bug report if necessary, and/or fix the problem forthwith.
(NB: If you are a system administrator and have no idea what this
message is talking about, this may indicate a serious mail system
misconfiguration somewhere. Please contact [email protected]
immediately.)
--
1093969: https://bugs.debian.org/cgi-bin/bugreport.cgi?bug=1093969
Debian Bug Tracking System
Contact [email protected] with problems
--- Begin Message ---
Source: mutter
Version: 47.4-1
User: [email protected]
Usertags: flaky
Severity: grave
Justification: blocks other packages from migrating
Dear Maintainers,
mutter autopkgtest are flaky and often fail and require multiple
manual reruns to let unrelated packages migrate, especially on
armhf/armel/ppc64el. As per RT, this kind of bug is RC.
Please consider either fixing those tests, or marking them with
Restrictions: flaky so that failures do not affect other packages.
https://ci.debian.net/packages/m/mutter/testing/armhf/
https://ci.debian.net/packages/m/mutter/testing/armel/
https://ci.debian.net/packages/m/mutter/testing/i386/
https://ci.debian.net/packages/m/mutter/testing/ppc64el/
--- End Message ---
--- Begin Message ---
Version: 48.3.1-1
On Fri, 24 Jan 2025 at 11:30:01 +0000, Luca Boccassi wrote:
mutter autopkgtest are flaky and often fail and require multiple
manual reruns to let unrelated packages migrate, especially on
armhf/armel/ppc64el.
The main root cause of test failures in mutter seems to have been
"In pixman_region32_init_rect: Invalid rectangle passed" (#1077800). We
are no longer seeing this failure mode, and I think that's because of
mutter!4039 upstream (48.3.1-1) or possibly some other upstream bug fix.
Another intermittent failure was "(EE) could not connect to wayland
server" which we have not seen since 48.2-1, so hopefully that is
another failure cause that has subsequently been fixed.
There have been no failures observed for release architectures on
ci.debian.net since 48.2-1 back in April, so I think whatever was the
root cause for the flaky autopkgtest must have been fixed - either fully
fixed, or at least, made substantially less likely.
(I'm not counting the more recent failure
https://ci.debian.net/packages/m/mutter/unstable/loong64/61423775/ on
loong64 as a problem with mutter itself, since it was an installability
issue during architecture bootstrapping rather than anything that can be
blamed on mutter.)
The most recent failure of the same test suite on the buildds (only
looking at release architectures) was 48.1-1 on amd64, which, again,
failed with symptoms matching #1077800. This appears to be fixed, or at
least a lot less likely, in newer upstream bugfix releases.
If similar test failures are still reproducible, please report a
separate bug describing the failure observed, with logs and whatever
details of the test system are available - we can't debug or fix a
failure that we can't observe. Ideally different failure modes should
have different bugs, similar to #1077800, #1050008 and so on, so that we
can keep track of what is fixed and what is still an issue.
For the build-time tests, any test failures will appear in the build log
as usual.
For the autopkgtests, currently the test log only provides a summary of
the test-cases that failed (for example in
https://ci.debian.net/packages/m/mutter/testing/armel/61424275/ the
single failure was mutter-16/surface-scale-tests.test) and the detailed
result is in the similarly-named file
mutter-16_surface-scale-tests.test.txt in the $AUTOPKGTEST_ARTIFACTS
(available on ci.debian.net as artifacts.tar.gz).
Thanks,
smcv
--- End Message ---
