• Type: Bug
• Resolution: Fixed
• Priority: Critical
• Component/s: gerrit-trigger-plugin
• Labels:

  • Aborting
  • gerrit
  • gerrit-trigger
  • plugin
• Environment:

  Jenkins version: 2.332
  gerrit-trigger version: 2.35.2
  Gerrit version: 3.5.0.1

[JENKINS-67761] New patchsets abort builds multiple times, preventing teardown actions

Sophia Bumgarner created issue - 2022-02-08 15:01

Sophia Bumgarner made changes - 2022-02-08 15:01

Description

Original: We have noticed that - for newly uploaded patch sets -the same build might get aborted multiple times. We assume that each build is aborted _n_ times if the corresponding changeset has triggered _n_ builds The second/third/... abortion may unfortunately act on an exit handler triggered by the initial (correct) abortion, cancelling the exit handler itself. In our case, this issue results in severe resource leakage, because essential teardown actions cannot run. This issue seemed to have have been solved by [PR#361|[https://github.com/jenkinsci/gerrit-trigger-plugin/pull/361],] unfortunately we are experiencing it yet again. An exemplary log might look like this: {code:java} 15:21:14 Aborted by new patch set. 15:21:14 Aborted by new patch set. 15:21:14 [Pipeline] // node 15:21:14 Aborted by new patch set. 15:21:14 [Pipeline] nodesByLabel 15:21:14 Found a total of 1 nodes with the '...' label 15:21:14 Aborted by new patch set. 15:21:14 [Pipeline] node 15:21:14 Aborted by new patch set. 15:21:14 Running on ... in /.../... 15:21:14 [Pipeline] // node 15:21:14 Aborted by new patch set. 15:21:14 Aborted by new patch set. 15:21:14 Aborted by new patch set. {code}

New: We have noticed that - for newly uploaded patch sets -the same build might get aborted multiple times. We assume that each build is aborted _n_ times if the corresponding changeset has triggered _n_ builds The second/third/... abortion may unfortunately act on an exit handler triggered by the initial (correct) abortion, cancelling the exit handler itself. In our case, this issue results in severe resource leakage, because essential teardown actions cannot run. This issue seemed to have have been solved by [PR#361|[https://github.com/jenkinsci/gerrit-trigger-plugin/pull/361]] unfortunately we are experiencing it yet again. An exemplary log might look like this: {code:java} 15:21:14 Aborted by new patch set. 15:21:14 Aborted by new patch set. 15:21:14 [Pipeline] // node 15:21:14 Aborted by new patch set. 15:21:14 [Pipeline] nodesByLabel 15:21:14 Found a total of 1 nodes with the '...' label 15:21:14 Aborted by new patch set. 15:21:14 [Pipeline] node 15:21:14 Aborted by new patch set. 15:21:14 Running on ... in /.../... 15:21:14 [Pipeline] // node 15:21:14 Aborted by new patch set. 15:21:14 Aborted by new patch set. 15:21:14 Aborted by new patch set. {code}

Sophia Bumgarner made changes - 2022-02-08 15:02

Description

Original: We have noticed that - for newly uploaded patch sets -the same build might get aborted multiple times. We assume that each build is aborted _n_ times if the corresponding changeset has triggered _n_ builds The second/third/... abortion may unfortunately act on an exit handler triggered by the initial (correct) abortion, cancelling the exit handler itself. In our case, this issue results in severe resource leakage, because essential teardown actions cannot run. This issue seemed to have have been solved by [PR#361|[https://github.com/jenkinsci/gerrit-trigger-plugin/pull/361]] unfortunately we are experiencing it yet again. An exemplary log might look like this: {code:java} 15:21:14 Aborted by new patch set. 15:21:14 Aborted by new patch set. 15:21:14 [Pipeline] // node 15:21:14 Aborted by new patch set. 15:21:14 [Pipeline] nodesByLabel 15:21:14 Found a total of 1 nodes with the '...' label 15:21:14 Aborted by new patch set. 15:21:14 [Pipeline] node 15:21:14 Aborted by new patch set. 15:21:14 Running on ... in /.../... 15:21:14 [Pipeline] // node 15:21:14 Aborted by new patch set. 15:21:14 Aborted by new patch set. 15:21:14 Aborted by new patch set. {code}

New: We have noticed that - for newly uploaded patch sets -the same build might get aborted multiple times. We assume that each build is aborted _n_ times if the corresponding changeset has triggered _n_ builds The second/third/... abortion may unfortunately act on an exit handler triggered by the initial (correct) abortion, cancelling the exit handler itself. In our case, this issue results in severe resource leakage, because essential teardown actions cannot run. This issue seemed to have have been solved by [PR361|[https://github.com/jenkinsci/gerrit-trigger-plugin/pull/361]] unfortunately we are experiencing it yet again. An exemplary log might look like this: {code:java} 15:21:14 Aborted by new patch set. 15:21:14 Aborted by new patch set. 15:21:14 [Pipeline] // node 15:21:14 Aborted by new patch set. 15:21:14 [Pipeline] nodesByLabel 15:21:14 Found a total of 1 nodes with the '...' label 15:21:14 Aborted by new patch set. 15:21:14 [Pipeline] node 15:21:14 Aborted by new patch set. 15:21:14 Running on ... in /.../... 15:21:14 [Pipeline] // node 15:21:14 Aborted by new patch set. 15:21:14 Aborted by new patch set. 15:21:14 Aborted by new patch set. {code}

Sophia Bumgarner made changes - 2022-02-08 15:03

Description

Original: We have noticed that - for newly uploaded patch sets -the same build might get aborted multiple times. We assume that each build is aborted _n_ times if the corresponding changeset has triggered _n_ builds The second/third/... abortion may unfortunately act on an exit handler triggered by the initial (correct) abortion, cancelling the exit handler itself. In our case, this issue results in severe resource leakage, because essential teardown actions cannot run. This issue seemed to have have been solved by [PR361|[https://github.com/jenkinsci/gerrit-trigger-plugin/pull/361]] unfortunately we are experiencing it yet again. An exemplary log might look like this: {code:java} 15:21:14 Aborted by new patch set. 15:21:14 Aborted by new patch set. 15:21:14 [Pipeline] // node 15:21:14 Aborted by new patch set. 15:21:14 [Pipeline] nodesByLabel 15:21:14 Found a total of 1 nodes with the '...' label 15:21:14 Aborted by new patch set. 15:21:14 [Pipeline] node 15:21:14 Aborted by new patch set. 15:21:14 Running on ... in /.../... 15:21:14 [Pipeline] // node 15:21:14 Aborted by new patch set. 15:21:14 Aborted by new patch set. 15:21:14 Aborted by new patch set. {code}

New: We have noticed that - for newly uploaded patch sets -the same build might get aborted multiple times. We assume that each build is aborted _n_ times if the corresponding changeset has triggered _n_ builds The second/third/... abortion may unfortunately act on an exit handler triggered by the initial (correct) abortion, cancelling the exit handler itself. In our case, this issue results in severe resource leakage, because essential teardown actions cannot run. This issue seemed to have have been solved by [PR#361|https://github.com/jenkinsci/gerrit-trigger-plugin/pull/361], unfortunately we are experiencing it yet again. An exemplary log might look like this: {noformat} 15:21:14 Aborted by new patch set. 15:21:14 Aborted by new patch set. 15:21:14 [Pipeline] // node 15:21:14 Aborted by new patch set. 15:21:14 [Pipeline] nodesByLabel 15:21:14 Found a total of 1 nodes with the '...' label 15:21:14 Aborted by new patch set. 15:21:14 [Pipeline] node 15:21:14 Aborted by new patch set. 15:21:14 Running on ... in /.../... 15:21:14 [Pipeline] // node 15:21:14 Aborted by new patch set. 15:21:14 Aborted by new patch set. 15:21:14 Aborted by new patch set. {noformat}

ckreisl made changes - 2022-11-12 13:53

Released As		New: Fixed in Gerrit Trigger vers. 2.37.0, by PR #478.
Resolution		New: Fixed [ 1 ]
Status	Original: Open [ 1 ]	New: Closed [ 6 ]