Uploaded image for project: 'Jenkins'
  1. Jenkins
  2. JENKINS-43635

declarative pipeline fails with workflow library

    XMLWordPrintable

    Details

    • Similar Issues:

      Description

      I have some multibranch pipelines, where I want to move over to the declarative pipeline syntax.
      I use a shared workflow library (added implicitly to all jobs, set as the default in the master config).
      When using declarative pipelines I get

      21:38:23 + docker-compose -f docker-compose.yml ps
      21:38:24 stty: standard input: Inappropriate ioctl for device
      21:38:24 Name Command State Ports 
      21:38:24 -----------------------------------------------------------------------------------------------------------------------------
      21:38:24 serverfeaturetacjfrsr167b43eenhq7fwkjpebbaprznkzesfair6fcv3a7pw26uh2fvuomq5a_oracle_1 /bin/sh -c /run.sh Up 1521/tcp 
      [Pipeline] sh
      21:38:24 [server_feature_TACJFRSR-167-B43EENHQ7FWKJPEBBAPRZNKZESFAIR6FCV3A7PW26UH2FVUOMQ5A] Running shell script
      [Pipeline] sh
      21:38:24 [server_feature_TACJFRSR-167-B43EENHQ7FWKJPEBBAPRZNKZESFAIR6FCV3A7PW26UH2FVUOMQ5A] Running shell script
      [Pipeline] }
      .....
      Caused by: java.lang.UnsupportedOperationException: Refusing to marshal org.codehaus.groovy.runtime.GStringImpl for security reasons
      

      when running the helper method from my pipeline library. The same helper method with the same arguments works just fine in a "classic" style Jenkinsfile pipeline. 
      So I headed off and put @NonCPS on the helper-method, the strange thing is that I then get another return type from that method! w/o annotation it's a List as expected, with annotation it becomes a String:

      java.lang.ClassCastException: org.jenkinsci.plugins.workflow.steps.EnvStep.overrides expects java.util.List<java.lang.String> but received class java.lang.String
       
      

      It seems strange that the same pipeline-library behaves differently between the two pipeline styles of declaration.
      See gist for full details:
      https://gist.github.com/davidkarlsen/882af8fa46859134aa3502fd5921cb79#file-gistfile1-txt-L5

        Attachments

          Issue Links

            Activity

            Hide
            abayer Andrew Bayer added a comment -

            Can you include the Jenkinsfile (or a dummy Jenkinsfile that reproduces the error) as well?

            Show
            abayer Andrew Bayer added a comment - Can you include the Jenkinsfile (or a dummy Jenkinsfile that reproduces the error) as well?
            Hide
            davidkarlsen davidkarlsen added a comment -

            Andrew Bayer attached.
            Maybe you need the workflow lib as well?

            Show
            davidkarlsen davidkarlsen added a comment - Andrew Bayer attached. Maybe you need the workflow lib as well?
            Hide
            abayer Andrew Bayer added a comment -

            You've got that in the gist, I think, but attaching it here wouldn't hurt.

            Show
            abayer Andrew Bayer added a comment - You've got that in the gist, I think, but attaching it here wouldn't hurt.
            Hide
            davidkarlsen davidkarlsen added a comment -
            Show
            davidkarlsen davidkarlsen added a comment - Andrew Bayer done
            Hide
            jglick Jesse Glick added a comment -

            The @NonCPS case is a red herring: JENKINS-31314. You may not mark this method @NonCPS.

            Show
            jglick Jesse Glick added a comment - The @NonCPS case is a red herring:  JENKINS-31314 . You may not mark this method @NonCPS .
            Hide
            davidkarlsen davidkarlsen added a comment -

            OK - but it failed without @NonCPS too...

            Show
            davidkarlsen davidkarlsen added a comment - OK - but it failed without @NonCPS too...
            Hide
            abayer Andrew Bayer added a comment -

            The @NonCPS failure is, like Jesse said, a red herring, in that it'd fail like that even if everything else was working fine. The non-@NonCPS case, however, is a legit bug independent of anything else.

            Show
            abayer Andrew Bayer added a comment - The @NonCPS failure is, like Jesse said, a red herring, in that it'd fail like that even if everything else was working fine. The non- @NonCPS case, however, is a legit bug independent of anything else.
            Hide
            abayer Andrew Bayer added a comment -

            davidkarlsen One little request - any chance you could post the full log? If not, I understand.

            Show
            abayer Andrew Bayer added a comment - davidkarlsen One little request - any chance you could post the full log? If not, I understand.
            Hide
            abayer Andrew Bayer added a comment - - edited

            And of course, because that'd be too easy, I can't reproduce this. Can you tell me what versions of the following plugins you have installed?

            • pipeline-model-definition
            • workflow-cps
            • workflow-api

            Thanks!

            Show
            abayer Andrew Bayer added a comment - - edited And of course, because that'd be too easy, I can't reproduce this. Can you tell me what versions of the following plugins you have installed? pipeline-model-definition workflow-cps workflow-api Thanks!
            Hide
            davidkarlsen davidkarlsen added a comment -

            pipeline model def: 1.0.2
            The other two I don't find (but they might have changed name?)
            Anyway I took a support bundle and the console log and sent in the cloudbees internal registered issue due to privacy reasons.

            Show
            davidkarlsen davidkarlsen added a comment - pipeline model def: 1.0.2 The other two I don't find (but they might have changed name?) Anyway I took a support bundle and the console log and sent in the cloudbees internal registered issue due to privacy reasons.
            Hide
            abayer Andrew Bayer added a comment -

            Ok, the actual issue is in the expression, and was fixed in 1.1 (JENKINS-42498).

            Show
            abayer Andrew Bayer added a comment - Ok, the actual issue is in the expression , and was fixed in 1.1 ( JENKINS-42498 ).
            Hide
            abayer Andrew Bayer added a comment -

            That is, it's this code:

                  when {
                    //if it is a feature branch - rebase in changes from develop branch - the branch we target integration into:
                    expression { isFeatureBranch(env) }
                  }
            
            Show
            abayer Andrew Bayer added a comment - That is, it's this code: when { // if it is a feature branch - rebase in changes from develop branch - the branch we target integration into: expression { isFeatureBranch(env) } }
            Hide
            jglick Jesse Glick added a comment -

            So should this be closed as a duplicate?

            Show
            jglick Jesse Glick added a comment - So should this be closed as a duplicate?
            Hide
            davidkarlsen davidkarlsen added a comment -

            Andrew Bayer How strange that it does not fail early on - as that expression is evaluated at the very start of the pipeline - long before we gets to the stage which use the global pipeline library part (which calls compose/shellscripts).

            Show
            davidkarlsen davidkarlsen added a comment - Andrew Bayer How strange that it does not fail early on - as that expression is evaluated at the very start of the pipeline - long before we gets to the stage which use the global pipeline library part (which calls compose/shellscripts).
            Hide
            davidkarlsen davidkarlsen added a comment -

            Thanks though!

            Show
            davidkarlsen davidkarlsen added a comment - Thanks though!
            Hide
            abayer Andrew Bayer added a comment -

            Yeah, the serialization errors can happen at pretty much any time in the run. Very annoying, but fixed. =)

            Show
            abayer Andrew Bayer added a comment - Yeah, the serialization errors can happen at pretty much any time in the run. Very annoying, but fixed. =)
            Hide
            bitwiseman Liam Newman added a comment -

            Bulk closing resolved issues.

            Show
            bitwiseman Liam Newman added a comment - Bulk closing resolved issues.

              People

              Assignee:
              abayer Andrew Bayer
              Reporter:
              davidkarlsen davidkarlsen
              Votes:
              0 Vote for this issue
              Watchers:
              5 Start watching this issue

                Dates

                Created:
                Updated:
                Resolved: