action #47219
closed[functional][u][easy] test fails in keymap_or_locale - rogue workqueue lockup
100%
Description
Observation¶
openQA test in scenario sle-15-SP1-Installer-DVD-aarch64-Build159.1-textmode@aarch64 fails in
keymap_or_locale
Reproducible¶
Looks sporadic
Expected result¶
Last good: 158.4 (or more recent)
Further details¶
There's a rogue workeque lockup present in the screen and serial0.txt log, but no further information about it
Suggestions¶
- - Write a new serial failure, so that these can be detected
- - Modify the post fail hook to collect at least dmesg and full journal log.
Updated by szarate about 6 years ago
Setting this to workable right away, I don't think should be complicated :)
Updated by szarate almost 6 years ago
- Subject changed from [functional][u] test fails in keymap_or_locale - rogue workqueue lockup to [functional][u][easy] test fails in keymap_or_locale - rogue workqueue lockup
Updated by okurz almost 6 years ago
- Target version changed from Milestone 22 to Milestone 23
yes, good idea.
Updated by jorauch almost 6 years ago
- Status changed from Workable to In Progress
According to the logs no post_fail_hook is being executed.
When setting the error type to 'hard' the test fails and we get the PFH of opensusebasetest.
However the big question is: How am I supposed to verify this?
Updated by jorauch almost 6 years ago
- Status changed from In Progress to Feedback
Updated by szarate almost 6 years ago
@jorauch for the time being I guess it's not required, as we know that this kind of errors are difficult to reproduce. Having the PFH sounds nice, however... in the case of this test it didn't cause any other test module to be tainted, could you use soft instead, so that those minutes are not wasted and further testing gets lost?
I thought for a moment that whenever a test that inherited from opensusebasetest, it would also inherit the PFH...
Updated by jorauch almost 6 years ago
If we fail hard but the test is not set to 'fatal' we should not lose any tests that follow?
Autoinst logs do not show any sign of PFH, but the screenshots seem to.
Now I am really confused
EDIT: also I think we should not softfail without bugref and if it appears in other modules the impact might be 'hard'
Updated by okurz almost 6 years ago
szarate wrote:
I thought for a moment that whenever a test that inherited from opensusebasetest, it would also inherit the PFH...
yes, that should be the case.
Updated by jorauch almost 6 years ago
- Status changed from Feedback to Resolved
PR merged, as its just a pattern detection of a random failure I would close this as we cannot be sure when we'll see it in production.
Updated by okurz almost 6 years ago
- Status changed from Resolved to Feedback
Please see https://github.com/os-autoinst/os-autoinst-distri-opensuse/pull/6851#issuecomment-467338617 . Either we should avoid the word "bug" or add a bug reference on our bugzilla instance.
Updated by jorauch almost 6 years ago
It's rogue, opening a bug will most likely have no effect, except we have a number to reference to.
Since the used module is named 'known_bugs' I would like to keep the naming and changing the commit message seems not feasible to me.
Updated by szarate almost 6 years ago
- Status changed from Feedback to Resolved
We can already detect the bug :), this ticket was for detecting the bug. I opened #48419 as a follow up, closing this one.
Updated by okurz almost 6 years ago
- Has duplicate action #48689: [functional][y] test fails in yast2_snapper - "rogue workqueue lockup bsc#1126782 - Serial error: BUG: workqueue lockup - pool cpus=0 node=0 flags=0x0 nice=0 stuck for 32s!" added
Updated by okurz almost 6 years ago
- Has duplicate deleted (action #48689: [functional][y] test fails in yast2_snapper - "rogue workqueue lockup bsc#1126782 - Serial error: BUG: workqueue lockup - pool cpus=0 node=0 flags=0x0 nice=0 stuck for 32s!")