action #135782
openauto-review+force-result ticket does not seem to have an effect when issue tracker changed after the initial comment when carry-over is effective
0%
Description
Observation¶
https://openqa.suse.de/tests/12082826 is failed, https://openqa.suse.de/tests/12082826#comments references a takeover ticket #134873 but the result is not updated when it should be.
ticket #134873 was moved to the tracker "openqa-force-result" in https://progress.opensuse.org/issues/134873#note-8 2023-09-06 10:29 (CEST?) and https://openqa.suse.de/tests/12082826 finished 2023-09-11 . The comment was carried over initially from https://openqa.suse.de/tests/12016780 which finished 2023-09-06 04:42:06 +0000 when the ticket was not yet in openqa-force-result.
Note: The job is softfailed now, but you can see that a human did that in the second comment.
Acceptance Criteria¶
- AC1: force result expressions in ticket titles are effective for comments with carried over labels even if the issue changed to be effective for force-result meanwhile
Suggestions¶
- Look for relevant logs from openqa-gru service or minion jobs https://openqa.suse.de/minion/jobs
- Confirm that it happens more often
- Maybe see if this is related to #135803 (just a guess, though)
- Since this seems to be a case where the tracker was wrong as the script was running:
- Consider adding a comment e.g. after "(Automatic takeover from t#12016780) (The hook script will not be executed.)" like "(force_result was specified but will not be used unless the tracker is $tracker)"
- If you expect that everything works as expected just find at least one openQA job where a label carry over and force-result was effective
Out of scope¶
- This is not about openqa-label-known-issues directly as the issue was about carried over comments where hook scripts are not executed as expected. See https://github.com/os-autoinst/scripts/blob/382c30836a5254515e474eab7ae36bf480e2286a/openqa-label-known-issues#L85
Updated by okurz 10 months ago
- Related to action #107923: qem-bot: Ignore not-ok openQA jobs for specific incident based on openQA job comment size:M added
Updated by tinita 10 months ago
Isn't that behaving like we currently would expect? Failures with takeover tickets don't go through the hook scripts at all, like the comment says:
(Automatic takeover from t#12016780) (The hook script will not be executed.)
Or is it a feature that I missed that now hook scripts should be executed also for takeover tickets?
Updated by okurz 10 months ago
tinita wrote in #note-4:
Isn't that behaving like we currently would expect? Failures with takeover tickets don't go through the hook scripts at all, like the comment says:
(Automatic takeover from t#12016780) (The hook script will not be executed.)
Or is it a feature that I missed that now hook scripts should be executed also for takeover tickets?
I understand https://github.com/os-autoinst/openQA/pull/5039 and as the commit says "This approach is mentioned in point 3.2 of https://progress.opensuse.org/issues/123724#note-12 as an alternative to executing the hook script when the carry-over happens." So my understanding is that here on carry-over the force-result should have an impact though no hook scripts should be called.
Updated by okurz 10 months ago
- Related to action #123724: auto_review not working despite ticket in openQA auto review project size:M added
Updated by tinita 10 months ago
I think the problem is that the original job was finished before the referenced ticket had the openqa-force-result
tracker assigned.
https://openqa.suse.de/tests/12016780#comments
geekotest wrote 2023-09-06 04:42:06 +0000
https://progress.opensuse.org/issues/134873#note-8
Tracker changed from action to openqa-force-result
2023-09-06 10:29
As a result, the hook script only added a normal comment. If it had had the tracker assigned at that time, then the hook script would have added a comment like label:force_result:softfailed:poo#134873 ticket-subject
.
Updated by okurz 10 months ago
- Subject changed from auto-review+force-result ticket does not seem to have an effect to auto-review+force-result ticket does not seem to have an effect when issue tracker changed after the initial comment when carry-over is effective
- Description updated (diff)
- Category changed from Regressions/Crashes to Feature requests
- Priority changed from High to Low
- Target version changed from Ready to future
In the team's estimation we further elaborated and found that the current behaviour is fine as in safe and conservative albeit with a certain caveat which we just need to accept for now. Ticket still valid though.