Uploaded image for project: 'Jenkins'
  1. Jenkins
  2. JENKINS-69338

Aborted build stays in queue after new gerrit patchset is created

XMLWordPrintable

      Patchset "A" is created in a gerrit review and a build is put in the Jenkins build queue. If patchset "B" is created before patchset "A" builds, the build for patchset "A" is aborted but it stays in the jenkins build queue and never clears out.  This can cause us to have over a 100 jobs queued up that will never build or be removed from the queue.

          [JENKINS-69338] Aborted build stays in queue after new gerrit patchset is created

          adhira deogade added a comment -

          We are seeing the exact same issue with the Gerrit-trigger plugin version 2.35.4 too.

          adhira deogade added a comment - We are seeing the exact same issue with the Gerrit-trigger plugin version 2.35.4 too.

          chencho chencho added a comment -

          Same here. 

          Versions:

          gerrit-trigger: 2.36.0

          jenkins: 2.346.3

          chencho chencho added a comment - Same here.  Versions: gerrit-trigger: 2.36.0 jenkins: 2.346.3

          Alex Z added a comment -

          +1 for issue.
          2.361.2 / 2.36.0

          Alex Z added a comment - +1 for issue. 2.361.2 / 2.36.0

          Kevin Cruise added a comment -

          Also +1 for issue

           2.332.3 / 2.36.0

          Kevin Cruise added a comment - Also +1 for issue  2.332.3 / 2.36.0

          ckreisl added a comment -

          Do you still see this kind of issue? I'll assume this has nothing todo with the plugin but the way how Jenkins itself handles the queue tasks. So far haven't seen this problem in latest Jenkins releases.

          ckreisl added a comment - Do you still see this kind of issue? I'll assume this has nothing todo with the plugin but the way how Jenkins itself handles the queue tasks. So far haven't seen this problem in latest Jenkins releases.

          Tomas added a comment - - edited

          We are experiencing the same problem (Jenkins 2.387.2), the repeated scenario is:

          commit chain A-B
          - change A (new patchset -> build)
          - change A again (new patchset -> build aborted, new build)
          - rebase B (new patchset -> build in queue, executes after A)
          - change B (new patchset -> build in queue - NO ABORT because of no build running on commit, executes after first build of B)

          Tomas added a comment - - edited We are experiencing the same problem (Jenkins 2.387.2), the repeated scenario is: commit chain A-B - change A (new patchset -> build) - change A again (new patchset -> build aborted, new build) - rebase B (new patchset -> build in queue, executes after A) - change B (new patchset -> build in queue - NO ABORT because of no build running on commit, executes after first build of B)

          Radoslaw Moszczynski added a comment -

          Same thing here on 2.387.3.

          Radoslaw Moszczynski added a comment - Same thing here on 2.387.3.

          Michael Nazzareno added a comment -

          Can you please try

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

          rmoszczy

          Michael Nazzareno added a comment - Can you please try https://github.com/jenkinsci/gerrit-trigger-plugin/pull/504 rmoszczy

          Robyn added a comment -

          I believe the initial issue is partially fixed, thank you!  The aborted builds no longer stay in the queue.  But now what happens is

          If patchset A is in the build queue, and patchset B is pushed. Both patchset A and B will be built. The B patchset does not abort/kill the queued A patchset.  

          This causes us to use extra jenkins agents and resources than what we should be using because we end up with unnecessary builds running.

          Robyn added a comment - I believe the initial issue is partially fixed, thank you!  The aborted builds no longer stay in the queue.  But now what happens is If patchset A is in the build queue, and patchset B is pushed. Both patchset A and B will be built. The B patchset does not abort/kill the queued A patchset.   This causes us to use extra jenkins agents and resources than what we should be using because we end up with unnecessary builds running.

            rsandell rsandell
            robynsndv72 Robyn Sandova
            Votes:
            3 Vote for this issue
            Watchers:
            11 Start watching this issue

              Created:
              Updated: