Does not look like a test issue to me. I understand that we do not have many people immediately jumping into fixing code for PackageKit related bugs in bugzilla so let's try to help with investigation.
Looking into the test steps we can see that packagekitd at least is accessed successfully during https://openqa.opensuse.org/tests/641616#step/updates_packagekit_gpk/16 even though "no updates" does not tell if packagekit was able to really query the repos. However, the first module trying to install packages which are not already installed is actually thunderbird, which does succeed, chromium is second and passing, and glxgears the third, also passing.
Looking into https://openqa.opensuse.org/tests/641616/file/inkscape-journal.log I can find some output regarding PackageKit, e.g. I see it being accessed what looks like our update test, then the daemon is quit - I think by the console tests - and later activated again - requested by gnome-software. Shortly before fwupd starts, also requested by gnome-software, and then core dumps. The core dumps are also included in https://openqa.opensuse.org/tests/641616/file/inkscape-problem_detection_logs.tar.xz :
Mar 23 16:27:46 susetest systemd[1]: Started PackageKit Daemon.
Mar 23 16:27:46 susetest fwupd[2651]: free(): invalid pointer
Mar 23 16:27:46 susetest systemd[1]: Started Process Core Dump (PID 2685/UID 0).
Mar 23 16:27:47 susetest systemd[1]: fwupd.service: Main process exited, code=dumped, status=6/ABRT
Mar 23 16:27:47 susetest systemd[1]: fwupd.service: Failed with result 'core-dump'.
Mar 23 16:27:47 susetest systemd[1]: Failed to start Firmware update daemon.
Mar 23 16:27:47 susetest systemd-coredump[2686]: Process 2651 (fwupd) of user 0 dumped core.
-> seems bug-worthy.
but it seems PackageKit is fine to search files:
Mar 23 16:27:48 susetest PackageKit[2659]: search-file transaction /1_babedcab from uid 1000 finished with success after 1527ms
Mar 23 16:27:48 susetest PackageKit[2659]: search-file transaction /2_dceedabb from uid 1000 finished with success after 564ms
Mar 23 16:27:49 susetest PackageKit[2659]: search-file transaction /3_abeddadd from uid 1000 finished with success after 535ms
Mar 23 16:27:49 susetest PackageKit[2659]: search-file transaction /4_eeddaaac from uid 1000 finished with success after 524ms
Mar 23 16:27:50 susetest PackageKit[2659]: search-file transaction /5_aadedbda from uid 1000 finished with success after 593ms
Mar 23 16:27:51 susetest PackageKit[2659]: search-file transaction /6_aaccacae from uid 1000 finished with success after 577ms
...
Mar 23 16:28:20 susetest PackageKit[2659]: search-file transaction /53_adaceabc from uid 1000 finished with success after 114ms
Mar 23 16:28:20 susetest polkitd[928]: Operator of unix-session:1 successfully authenticated as unix-user:root to gain TEMPORARY authorization for action org.freedesktop.packagekit.package-install for system-bus-name::1.69 [pkcon install -yp inkscape] (owned by unix-user:bernhard)
Mar 23 16:28:20 susetest PackageKit[2659]: new install-packages transaction /46_aabaaaba scheduled from uid 1000
Mar 23 16:28:20 susetest PackageKit[2659]: uid 1000 obtained auth for org.freedesktop.packagekit.package-install
but then suddenly
Mar 23 16:28:26 susetest PackageKit[2659]: search-file transaction /98_bcabcddb from uid 1000 finished with success after 133ms
Mar 23 16:28:26 susetest PackageKit[2659]: get-details transaction /99_ededceea from uid 1000 finished with success after 53ms
Mar 23 16:28:46 susetest PackageKit[2659]: daemon quit
Mar 23 16:28:46 susetest packagekitd[2659]: Source ID 16 was not found when attempting to remove it
I am not sure if this is a report about a successful operation but I wonder why the "daemon quit" appears.
I did not find anything else useful.
Suggestions how to continue¶
- Report bug about fwupd core-dumping
- include file /var/log/pk_backend_zypp whenever pkcon is involved. I am thinking about a member variable to the test object which is evaluated in the post_fail_hook, e.g.
$self->{include_pk_backend_zypp_log} = 1;
- Crosscheck other failed jobs as well as not-failed jobs - or failed jobs that failed in a later module not related to pk to have same logs recorded but covering the packagekit related tests. Check if the same errors appear in the bad+good jobs