Issue

A branch indexing scan job for a multibranch pipeline project has been running for a long time. The job does not respond to abort requests via the Jenkins UI. This is preventing new branches from being discovered. Users are forced to restart their Jenkins instance to cancel the job.

Environment

• CloudBees CI (CloudBees Core)
• CloudBees Jenkins Enterprise
• CloudBees Jenkins Platform - Client controller
• Jenkins LTS
• Pipeline: Multibranch

Workaround

This issue was reported in JENKINS-48933. Please contact CloudBees Support so they can collect the necessary data to determine root cause.

As a workaround, the following groovy script can be used to abort branch indexing jobs that have been running for a specified time period.

import groovy.time.TimeCategory

/**
 * Find all branch indexing that have been running for more than the **delay** specified and interrupt them.
 */
use(TimeCategory) {
  def delay = 2.hours
  jenkins.model.Jenkins.instanceOrNull.toComputer().allExecutors
    .findAll { exec -> 
      exec.elapsedTime > delay.toMilliseconds()
      exec.currentExecutable != null && 
      exec.currentExecutable instanceof jenkins.branch.MultiBranchProject.BranchIndexing 
    }.each { exec ->
      println " * Stopping Branch Indexing of ${exec.currentExecutable.parent.fullDisplayName} that spent ${exec.elapsedTime}ms scanning on ${exec.owner.displayName} #${exec.number}..."
      exec.interrupt(hudson.model.Result.ABORTED, new jenkins.model.CauseOfInterruption() {
        @Override
        String getShortDescription() {
          return "Branch Indexing was running for too long. See Jenkins administrators"
        }
      })
    }
}

return