QA » openQA Project » openQA Tests

build was obsoleted, retriggered some jobs in the same build.

And now of course tests were not working anymore because the incident repo vanished. Retrying with more recent one:

for i in {001..100} ; do openqa_clone_job_o3 --skip-chained-deps 710696 TEST=okurz_poo38387_$i BUILD=20180721-3:poo38387 _GROUP="Development Leap" ; done

-> https://openqa.opensuse.org/tests/overview?groupid=39&distri=opensuse&version=15.0&build=20180721-3%3Apoo38387

again obsoleted by more recent update tests. This is not going to work this way. Nevertheless, taking a look at most recent production runs I could find https://openqa.opensuse.org/tests/711279#step/gnucash/22 where the "window" menu is activated after the dialog to "save file changes" is discarded. Looks like "alt-w" can cause this.

https://openqa.opensuse.org/tests/711074#step/gnucash/20 is another failure, openSUSE Tumbleweed gnome scenario, which shows the clock time 03:14 before start and also when seemingly the screen lock shows up, unless this is actually caused by the key presses?

Tried to test for stability locally again and hit another interesting/stupid/annoying/funny problem: http://lord.arch/tests/1729#step/gnucash/15 shows that the screen locked in the same minute that we installed packages, which is just one minute after we did the last typing so there should be no screenlock and not screen saver. However, taking a look at the journal in http://lord.arch/tests/1729/file/gnucash-journal.log one can see a sudden huge 3-day jump in the timestamps together with according output from chrony:

Jul 20 10:20:54 susetest PackageKit[18076]: install-packages transaction /4_edcceeaa from uid 1000 finished with success after 55100ms
Jul 20 10:20:54 susetest polkitd[838]: Unregistered Authentication Agent for unix-process:18082:168414 (system bus name :1.134, object path /org/freedesktop/PolicyKit1/AuthenticationAgent, locale en_US.UTF-8) (disconnected from bus)
Jul 20 10:20:56 susetest PackageKit[18076]: get-updates transaction /5_abcccacd from uid 1000 finished with success after 1471ms
Jul 20 10:20:57 susetest chronyd[713]: Selected source 217.114.59.3
Jul 20 10:20:57 susetest chronyd[713]: System clock wrong by 285436.951742 seconds, adjustment started
Jul 23 17:38:14 susetest chronyd[713]: System clock was stepped by 285436.951741 seconds
Jul 23 17:38:16 susetest chronyd[713]: Source 139.162.219.252 replaced with 178.79.160.57
Jul 23 17:38:33 susetest PackageKit[18076]: daemon quit
Jul 23 17:38:33 susetest packagekitd[18076]: Source ID 16 was not found when attempting to remove it
Jul 23 17:38:44 susetest unix2_chkpwd[20121]: pam_unix(xscreensaver:auth): authentication failure; logname= uid=1000 euid=0 tty= ruser= rhost=  user=bernhard
Jul 23 17:38:46 susetest unix2_chkpwd[20121]: pam_authenticate(xscreensaver, bernhard): Authentication failure

so what happened is that the test which I started from an older qcow snapshot for faster reproduction of good statistics ended up updating the time to the current time causing the screenlock to suddenly trigger which is certainly totally unexpected by the test. To ensure the test continuing in this very special – development approach specific – scenario would be disable the screensaver on lxde (we already do something similar on gnome and plasma). E.g. see https://unix.stackexchange.com/questions/356548/how-to-disable-screensaver-in-lxde and https://www.linuxquestions.org/questions/linux-desktop-74/disable-screensaver-in-lxde-745203/ or simply kill xscreensaver or something for the current session. Or I just reconduct the test from x11-user_gui_login I guess. What I tried for now is a local change

 sub handle_relogin {
     handle_logout;
+    assert_screen 'displaymanager', 90;
+    select_console 'root-console';
+    assert_script_run 'chronyc makestep';
+    select_console 'x11', await_console => 0;
     handle_login;
 }

20/20 jobs passed :(

Rebased test code and triggered run 21..60

Finally managed to reproduce some failures again: http://lord.arch/tests/overview?build=20180724:1&distri=opensuse&version=Tumbleweed shows now 6/60 jobs. I can identify one VLC needle that needs to be fixed and chromium failures. I guess it makes sense if I continue here because the current problem is not that I see how I could work any faster or need any help but just need to wait for long-running test jobs.

It may be a race condition on sending keys:

• check-/assert_screen before sending key (sync point)

1. alt-f4 is pressed to close the help window
2. alt-c is pressed to close the tips window
3. ctrl-q is pressed to close gnucash

The problem is that sometimes ctrl-q seems to not take effect.
Trying asserting the screen before sending ctrl-q.

At the moment I need to figure out how to get this file on the worker /usr/share/qemu/ovmf-x86_64-ms-code.bin so I can verify locally (package qemu-ovmf).

All failed in "zypper_add_repos" as the maintenance repo vanished in the mean time.

Let's try based on the latest aggregated repo:

$ for i in {001..100} ; do openqa_clone_job_o3 735474 TEST=sergio_poo38387_$i BUILD=20180817-3:poo38387 _GROUP="Development Leap" ; done

-> https://openqa.opensuse.org/tests/overview?version=15.0&build=20180817-3%3Apoo38387&groupid=39&distri=opensuse

Used as label in https://openqa.opensuse.org/tests/739102#comments then, thx

Automatic suspend is a different and sporadic issue that will be handled here: https://progress.opensuse.org/issues/40172

PR: https://github.com/os-autoinst/os-autoinst-needles-opensuse/pull/424

New needle for Leap (font type differs from tumbleweed): test-gnucash-1-tip-gnome-leap-20180823

Waiting for verification run on OSD:

• https://openqa.opensuse.org/tests/741002
• https://openqa.opensuse.org/tests/741003
• https://openqa.opensuse.org/tests/741004

Unfortunately, still fails on Leap 42.3: https://openqa.opensuse.org/tests/741153#step/gnucash/21.
Leap 15 is fine.

That job started yesterday at 8:00~, my needle was created on the afternoon. My verification runs were canceled and the OBS repo removed. I am having problems to verify...

ok, now it was about Leap 42.3, which also has a slightly different font type: test-gnucash-1-tip-gnome-leap-42.3-20180824

Waiting for verification run on O3: https://openqa.opensuse.org/tests/741205