Uploaded image for project: 'Jenkins'
  1. Jenkins
  2. JENKINS-43339

`currentBuild.result` is not set to `ABORTED` when the build was aborted

      Right now when the build is aborted using the `pipeline-input-step` plugin in a declarative pipeline, the build result is set to `FAILURE` instead of `ABORTED` in the post section. However, the actual result is properly set to `ABORTED` afterwards. This makes it impossible to do conditional code in the post section.

          [JENKINS-43339] `currentBuild.result` is not set to `ABORTED` when the build was aborted

          Roch Devost added a comment -

          It seems this is also true for `SUCCESS` which is set only after the build has completed (and after all post sections), but using `currentBuild.result` inside of a post section returns `null`.

          Roch Devost added a comment - It seems this is also true for `SUCCESS` which is set only after the build has completed (and after all post sections), but using `currentBuild.result` inside of a post section returns `null`.

          Jo Carter added a comment -

          This also happens for builds that are Superseded (when using milestones).  When I get the current result status in the failure block it says "FAILURE" but the build final status is NOT_BUILT. (Same for ABORTED as above).

          It's also incorrectly reporting as a completeBuild - which is an additional check that I'm performing.

          Getting the previous build does return the correct result.

          Jo Carter added a comment - This also happens for builds that are Superseded (when using milestones).  When I get the current result status in the failure block it says "FAILURE" but the build final status is NOT_BUILT. (Same for ABORTED as above). It's also incorrectly reporting as a completeBuild - which is an additional check that I'm performing. Getting the previous build does return the correct result.

          Andrew Bayer added a comment -

          Interesting! We're setting the build result to FAILURE explicitly when we encounter an exception in execution of a stage, but I'm guessing that's not the ideal behavior for these scenarios. I'll investigate.

          Andrew Bayer added a comment - Interesting! We're setting the build result to FAILURE explicitly when we encounter an exception in execution of a stage, but I'm guessing that's not the ideal behavior for these scenarios. I'll investigate.

          Jo Carter added a comment -

          Thank you

          Jo Carter added a comment - Thank you

          Andrew Bayer added a comment -

          Andrew Bayer added a comment - Tada - PR up at https://github.com/jenkinsci/pipeline-model-definition-plugin/pull/153 .

          Code changed in jenkins
          User: Andrew Bayer
          Path:
          pipeline-model-definition/pom.xml
          pipeline-model-definition/src/main/groovy/org/jenkinsci/plugins/pipeline/modeldefinition/Utils.groovy
          pipeline-model-definition/src/main/groovy/org/jenkinsci/plugins/pipeline/modeldefinition/model/conditions/NotBuilt.groovy
          pipeline-model-definition/src/main/resources/org/jenkinsci/plugins/pipeline/modeldefinition/ModelInterpreter.groovy
          pipeline-model-definition/src/main/resources/org/jenkinsci/plugins/pipeline/modeldefinition/agent/impl/DockerPipelineFromDockerfileScript.groovy
          pipeline-model-definition/src/main/resources/org/jenkinsci/plugins/pipeline/modeldefinition/agent/impl/DockerPipelineScript.groovy
          pipeline-model-definition/src/main/resources/org/jenkinsci/plugins/pipeline/modeldefinition/agent/impl/LabelScript.groovy
          pipeline-model-definition/src/main/resources/org/jenkinsci/plugins/pipeline/modeldefinition/model/conditions/Messages.properties
          pipeline-model-definition/src/test/java/org/jenkinsci/plugins/pipeline/modeldefinition/BasicModelDefTest.java
          pom.xml
          http://jenkins-ci.org/commit/pipeline-model-definition-plugin/d9a8fb10d9c830470dfb9871696ca9df8c50f0f7
          Log:
          [FIXED JENKINS-43339] Properly set result from FlowInterruptedException

          We should be setting the build result to
          FlowInterruptedException.getResult() when we catch a
          FlowInterruptedException, since there's actually a relevant result
          there. Also added a notBuilt post status while I was here.

          SCM/JIRA link daemon added a comment - Code changed in jenkins User: Andrew Bayer Path: pipeline-model-definition/pom.xml pipeline-model-definition/src/main/groovy/org/jenkinsci/plugins/pipeline/modeldefinition/Utils.groovy pipeline-model-definition/src/main/groovy/org/jenkinsci/plugins/pipeline/modeldefinition/model/conditions/NotBuilt.groovy pipeline-model-definition/src/main/resources/org/jenkinsci/plugins/pipeline/modeldefinition/ModelInterpreter.groovy pipeline-model-definition/src/main/resources/org/jenkinsci/plugins/pipeline/modeldefinition/agent/impl/DockerPipelineFromDockerfileScript.groovy pipeline-model-definition/src/main/resources/org/jenkinsci/plugins/pipeline/modeldefinition/agent/impl/DockerPipelineScript.groovy pipeline-model-definition/src/main/resources/org/jenkinsci/plugins/pipeline/modeldefinition/agent/impl/LabelScript.groovy pipeline-model-definition/src/main/resources/org/jenkinsci/plugins/pipeline/modeldefinition/model/conditions/Messages.properties pipeline-model-definition/src/test/java/org/jenkinsci/plugins/pipeline/modeldefinition/BasicModelDefTest.java pom.xml http://jenkins-ci.org/commit/pipeline-model-definition-plugin/d9a8fb10d9c830470dfb9871696ca9df8c50f0f7 Log: [FIXED JENKINS-43339] Properly set result from FlowInterruptedException We should be setting the build result to FlowInterruptedException.getResult() when we catch a FlowInterruptedException, since there's actually a relevant result there. Also added a notBuilt post status while I was here.

          Andrew Bayer added a comment -

          Released in 1.1.4 just now!

          Andrew Bayer added a comment - Released in 1.1.4 just now!

          Roman Pickl added a comment -

          Roman Pickl added a comment - I looked forward to this change, but my related problem is not fixed http://stackoverflow.com/questions/42973501/declarative-pipeline-donts-send-post-failure-notificaitons-if-build-was-abbort

          romanp I've been having the same problem as you and for me the issue you describe in stackoverflow was resolved with 1.1.4.

          Robin Björklin added a comment - romanp I've been having the same problem as you and for me the issue you describe in stackoverflow was resolved with 1.1.4.

          Roman Pickl added a comment -

          seems working again after restarting jenkins (maybe I've missed this after installing).
          sorry for any inconvenience.

          Roman Pickl added a comment - seems working again after restarting jenkins (maybe I've missed this after installing). sorry for any inconvenience.

          This issue still persist for us. We have script e.g.

          node {
              def repoGitUrl = "git@usfr-git.foo.git"
          
              jdk "JDK8"
          
              catchError {
                  dir('FOO') {
                      stage "Checkout"
                      gitCheckout(repoGitUrl, "master")
          
                      stage "Merge"
                      def gitflow = new com.foo.GitFlow()
                      gitflow.mergeMasterToOtherBranches()
          
                      deleteDir()
                  }
          
                  currentBuild.result = 'SUCCESS'
              }
          
              mailOnFailure("foo@bar.com")
          }
          
          The gitCheckout() function:
          def call(gitlab, branch) {
              checkout([
                      $class                           : 'GitSCM',
                      branches                         : [[name: "${branch}"]],
                      doGenerateSubmoduleConfigurations: false,
                      extensions                       : [
                              [$class: 'WipeWorkspace'],
                              [$class: 'CleanBeforeCheckout'],
                              [$class: 'LocalBranch', localBranch: "${branch}"]
                      ],
                      submoduleCfg                     : [],
                      userRemoteConfigs                : [
                              [
                                      credentialsId: 'GitLabBuildUser',
                                      url          : "${gitlab}"
                              ]
                      ]
              ])
          }
          
          The mailOnFailure() function:
          def call(recipients) {
              if (currentBuild.result != 'ABORTED') { // don't send email when the build was aborted
                  step([$class                  : 'Mailer',
                        notifyEveryUnstableBuild: true,
                        recipients              : recipients,
                        sendToIndividuals       : true])
              }
          }
          

           

          When I abort the Jenkins job while it is checking out, the currentBuild.result is FAILURE, not ABORTED as expected. Git checkout throws a java.lang.InterruptedException and then the currentBuild.result contains FAILURE, not ABORTED.

          Full log: https://pastebin.com/B5AFZnN9

           

          Versions:

          pipeline-model-definition 1.1.9
          pipeline-input-step 2.7
          Jenkins 2.46.3

          Antonín Karola added a comment - This issue still persist for us. We have script e.g. node { def repoGitUrl = "git@usfr-git.foo.git" jdk "JDK8" catchError { dir( 'FOO' ) { stage "Checkout" gitCheckout(repoGitUrl, "master" ) stage "Merge" def gitflow = new com.foo.GitFlow() gitflow.mergeMasterToOtherBranches() deleteDir() } currentBuild.result = 'SUCCESS' } mailOnFailure( "foo@bar.com" ) } The gitCheckout() function: def call(gitlab, branch) { checkout([ $class : 'GitSCM' , branches : [[name: "${branch}" ]], doGenerateSubmoduleConfigurations: false , extensions : [ [$class: 'WipeWorkspace' ], [$class: 'CleanBeforeCheckout' ], [$class: 'LocalBranch' , localBranch: "${branch}" ] ], submoduleCfg : [], userRemoteConfigs : [ [ credentialsId: 'GitLabBuildUser' , url : "${gitlab}" ] ] ]) } The mailOnFailure() function: def call(recipients) { if (currentBuild.result != 'ABORTED' ) { // don't send email when the build was aborted step([$class : 'Mailer' , notifyEveryUnstableBuild: true , recipients : recipients, sendToIndividuals : true ]) } }   When I abort the Jenkins job while it is checking out, the currentBuild.result is FAILURE, not ABORTED as expected. Git checkout throws a java.lang.InterruptedException and then the currentBuild.result contains FAILURE, not ABORTED. Full log: https://pastebin.com/B5AFZnN9   Versions: pipeline-model-definition 1.1.9 pipeline-input-step 2.7 Jenkins 2.46.3

          This happens for me right now as well with Jenkins 2.72.3. 

          Phillip Johnston added a comment - This happens for me right now as well with Jenkins 2.72.3. 

          Peter Wilcock added a comment -

          Also confirming this error with declarative pipelines. When I abort the build the post-build action for failure triggers, not abort.  

          Final two lines of console log show:

          ERROR: script returned exit code 143
          Finished: ABORTED
          

          Jenkins 2.93. 
          Latest pipeline plugins all installed as of this date. 

          Peter Wilcock added a comment - Also confirming this error with declarative pipelines. When I abort the build the post-build action for failure triggers, not abort.   Final two lines of console log show: ERROR: script returned exit code 143 Finished: ABORTED Jenkins 2.93.  Latest pipeline plugins all installed as of this date. 

          Andrew Bayer added a comment -

          Regrettably, not much we can do here - the ABORTED status, in some cases, doesn't get set until the Pipeline finishes executing, so we can't tell from Declarative that we should be treating the build as ABORTED.

          Andrew Bayer added a comment - Regrettably, not much we can do here - the ABORTED status, in some cases, doesn't get set until the Pipeline finishes executing, so we can't tell from Declarative that we should be treating the build as ABORTED .

          abayer, rather than just saying this won't be fixed, perhaps it is better to remove features that provide the illusion that `ABORTED` is a supported state that can be handled in declarative pipelines? Then there will be no confusion

          Phillip Johnston added a comment - abayer , rather than just saying this won't be fixed, perhaps it is better to remove features that provide the illusion that `ABORTED` is a supported state that can be handled in declarative pipelines? Then there will be no confusion

          (I suspect this issue will continue to be reported when someone else tries to handle `aborted` in a `post` block)

          Phillip Johnston added a comment - (I suspect this issue will continue to be reported when someone else tries to handle `aborted` in a `post` block)

          Andrew Bayer added a comment -

          The thing is that ABORTED does work in some contexts, but not others. It's weird.

          Andrew Bayer added a comment - The thing is that ABORTED does work in some contexts, but not others. It's weird.

          Peter Wilcock added a comment - - edited

          abayer I've observed the following:

          When running parallel steps with fastFail set, the fast-failed aborted build will return FAILURE if the act of aborting a step that was still in progress returns an error code. 

          However if other parallel steps have finished successfully, and the last running step fails, this correctly returns ABORTED in the event of failure.

          Some kind of race condition where if a step fails, fastFail aborts all other steps, and the last error code returned is the status seen by the declarative pipeline, but isn't the one respected by the overall build.status? Hopefully that makes sense. 

          Peter Wilcock added a comment - - edited abayer I've observed the following: When running parallel steps with fastFail set, the fast-failed aborted build will return FAILURE if the act of aborting a step that was still in progress  returns an error code.  However if other parallel steps have finished successfully, and the last running step fails, this correctly returns ABORTED in the event of failure. Some kind of race condition where if a step fails, fastFail aborts all other steps, and the last error code returned is the status seen by the declarative pipeline, but isn't the one respected by the overall build.status? Hopefully that makes sense. 

          kutzi added a comment - - edited

          Is there not a reliable way to get from the rawBuild if a build was aborted?
          I cannot image that there's no way - as the builds are eventually - when they are totally finished - always displayed with the right result. 'Only' currentBuild.currentResult is very often giving a different result (FAILURE usually)
          We're using this hack now to get better detection of aborted builds:

          if (currentBuild.rawBuild.getActions(jenkins.model.InterruptedBuildAction.class).isEmpty()) {
             return currentBuild.currentResult
          } else {
            return "ABORTED"
          }
          

          kutzi added a comment - - edited Is there not a reliable way to get from the rawBuild if a build was aborted? I cannot image that there's no way - as the builds are eventually - when they are totally finished - always displayed with the right result. 'Only' currentBuild.currentResult is very often giving a different result (FAILURE usually) We're using this hack now to get better detection of aborted builds: if (currentBuild.rawBuild.getActions(jenkins.model.InterruptedBuildAction.class).isEmpty()) { return currentBuild.currentResult } else { return "ABORTED" }

          Liam Newman added a comment -

          Bulk closing resolved issues.

          Liam Newman added a comment - Bulk closing resolved issues.

          Thanks kutzi – this workaround is indeed very helpful! (Of course one would prefer a proper/clean solution/fix... )

          Reinhold Füreder added a comment - Thanks kutzi – this workaround is indeed very helpful! (Of course one would prefer a proper/clean solution/fix... )

          kutzi added a comment - - edited

          reinholdfuereder according to your comment, you still have experienced this issue in the current release?
          I'm having the workaround in place and therefore not watching it anymore.
          According to Liam's comment this should have been fixed?

          kutzi added a comment - - edited reinholdfuereder  according to your comment, you still have experienced this issue in the current release? I'm having the workaround in place and therefore not watching it anymore. According to Liam's comment this should have been fixed?

          Aleksejs added a comment -

          Jenkins 2.375.1, manually aborted build had `currentBuild.currentResult`= success, sending notification as such. Workaround helped. 

          Aleksejs added a comment - Jenkins 2.375.1, manually aborted build had `currentBuild.currentResult`= success, sending notification as such. Workaround helped. 

            abayer Andrew Bayer
            rochdev Roch Devost
            Votes:
            6 Vote for this issue
            Watchers:
            27 Start watching this issue

              Created:
              Updated:
              Resolved: