-
Bug
-
Resolution: Won't Fix
-
Major
-
None
We have a job that uses fan-in to package up a final artifact after a number of other jobs have finished. It seems to work great, as long as the jobs involved are standard projects, not pipelines. If one of the jobs (or its ancestor) is a pipeline job, i get an exception:
00:00:41.665 FATAL: class org.jenkinsci.plugins.workflow.job.WorkflowJob cannot be cast to class hudson.model.AbstractProject (org.jenkinsci.plugins.workflow.job.WorkflowJob is in unnamed module of loader jenkins.util.URLClassLoader2 @46750bd7; hudson.model.AbstractProject is in unnamed module of loader org.eclipse.jetty.webapp.WebAppClassLoader @76f7d241) 00:00:41.665 java.lang.ClassCastException: class org.jenkinsci.plugins.workflow.job.WorkflowJob cannot be cast to class hudson.model.AbstractProject (org.jenkinsci.plugins.workflow.job.WorkflowJob is in unnamed module of loader jenkins.util.URLClassLoader2 @46750bd7; hudson.model.AbstractProject is in unnamed module of loader org.eclipse.jetty.webapp.WebAppClassLoader @76f7d241) 00:00:41.665 at org.lonkar.jobfanin.FanInReverseBuildTrigger.thisAndUpstreamAreNotBuildingAndStable(FanInReverseBuildTrigger.java:205) 00:00:41.665 at org.lonkar.jobfanin.FanInReverseBuildTrigger.allUpsteamIsBuild(FanInReverseBuildTrigger.java:186) 00:00:41.665 at org.lonkar.jobfanin.FanInReverseBuildTrigger.access$300(FanInReverseBuildTrigger.java:62) 00:00:41.665 at org.lonkar.jobfanin.FanInReverseBuildTrigger$1.shouldTriggerBuild(FanInReverseBuildTrigger.java:249) 00:00:41.665 at hudson.tasks.BuildTrigger.execute(BuildTrigger.java:291) 00:00:41.665 at hudson.model.AbstractBuild$AbstractBuildExecution.cleanUp(AbstractBuild.java:728) 00:00:41.665 at hudson.model.Build$BuildExecution.cleanUp(Build.java:194) 00:00:41.665 at hudson.model.Run.execute(Run.java:1947) 00:00:41.665 at hudson.model.FreeStyleBuild.run(FreeStyleBuild.java:44) 00:00:41.665 at hudson.model.ResourceController.execute(ResourceController.java:107) 00:00:41.665 at hudson.model.Executor.run(Executor.java:449)
Worse, this seems to stop the trigger steps. For example, if job A runs, there may be 4 or 5 jobs that are watching "A". If the 2nd of those is a job that uses fan-in, and it has an upstream job that is a pipeline, jobs 3-5 don't get triggered.
