rsandell seems that this not matching anymore cancelMatchingJobs jobName. Before NULL was passed, do you have some idea?

Hello ondrej_pik, panicking possible to provide your test scenario? Especially I'm interested in the the job settings and its Gerrit event triggers which should trigger the build.

Hello ckreisl , I am not sure what is the best way to share it. But the job is really simple. I created a separate job just for testing this problem.

It's pipeline project with only gerrit trigger set as shown on attached screenshots. And the Gerrit Server configuration is set like this:

The pipeline script is just

node() {
    sh "env"
    sh "sleep 600"
    echo "done"
}

The Gerrit changes (and the following new patchsets) were pushed directly to code review - not WIP/private/... .

Let me know if I can provide some other useful information.

Hi ondrej_pik, finally had some time to check on this and right now I can't reproduce this problem in my test environment . I used the same Jenkins Gerrit server settings and Gerrit event job settings with the following setup.

Jenkins vers. 2.346.3
Gerrit 3.7

What I did. The setup:

1. Created a Jenkins pipeline job called "verify-a".
2. Set the Gerrit event trigger with "Patchset created" listening on a dummy repository "repo-a".
3. Kept the job disabled for now to prepare the Gerrit patchsets
4. Uploaded first patch to repo-a (1)
5. Set the topic of created patchset (1) to "topicTest"
6. Created a second change on repo-a and uploaded it (2)
7. Set the same topic "topicTest" to the second patchset (2)
8. Enabled the "verify-a" job.

Test scenario:
1. Updated patchset (1) with a new commit which triggers "verify-a" with the correct Gerrit event parameters based on the event patchset created.
2. Updated patchset (2) with a new commit, the triggered event successfully aborts the previously triggered "verify-a" job based on patchset (1) and triggers a new one based on the new event from patchset (2).

//Edit. The branch for the two created patchsets was always "main" in this case.

Anything i missed here?

ckreisl Try to upload a new version of the same patchset so the old one should abort but nothing is cancelled

ckreisl git bisect come here

commit df491177aa13f60dbcdf2fb974af11c27c2295a0
Author: ckreisl <christoph.kreisl@gmx.de>
Date: Mon Oct 24 16:44:21 2022 +0200

Fix leftover of "Stopped" jobs in queue

I have a topic with 4 patches so I need to run as whole topic and build the last commit.

https://github.com/jenkinsci/gerrit-trigger-plugin/pull/504