Uploaded image for project: 'Jenkins'
  1. Jenkins
  2. JENKINS-54977

Failed one of sub-jobs in a phase, but the build went on despite a Success or Unstable setting

    • Icon: Bug Bug
    • Resolution: Unresolved
    • Icon: Minor Minor
    • multijob-plugin
    • None

      We have a multi-phase legacy job for product testing, where one phase has several sequential steps to set up a deployment, then another runs the tests against it, another analyzes the results etc.

      The sequential set-up phase in a recent build failed to install the product image, and as that sub-job in the phase had settings to fail the phase and kill other builds, other steps such as configuring that deployment indeed did not happen.

      However, even though this phase overall had a setting to continue into the next phase if the build is "Successful or Unstable, but not Failed", tests commenced against the incorrectly prepared product.

      The expected outcome was for the MultiPhase job to fail and not do any other steps right there, when it decided the installation was broken.

      We are now reconfiguring such jobs to require a "Successful" outcome (and hope it will behave better), but still suspect something can be wrong in the plugin. Or our setup and use of it, somehow...

          [JENKINS-54977] Failed one of sub-jobs in a phase, but the build went on despite a Success or Unstable setting

          John Rocha added a comment -

          I recently had the same problem. Where a multijob phase failed and was set to only continue if the phase was successful. However it continued even though the phase failed.

          This does not happen all the time. But it happens enough to cause problems.

          More details for our problem:

          1. Multi-job phase (unnamed)
            This job is set to sequential and to only continue when jobs' status are Successful
            1. sub-job#1: Run matrix job to build 32B and 64B on various machines
              Kill the phase on: Failure
            2. sub-job#2: Combine all of the successful build results into a build artifact and deploy to fileshare
              Kill the phase on Failure
          2. Multi-job phase (unnamed)
            Run jobs in parallel. Continue if jobs are stable or unstable but not failed
            1. Run matrix job to test 32B and 64B on various Windows OS versions
              Never kill this phase
          3. Multi-job phase (unnamed)
            Run jobs in parallel. Continue if jobs are successful
            1. Run jobs to deploy from "to-be-tested" fileshare location to production.
              Kill the phase on failure

          We had a failure during the first phase when the artifacts were being merged. The job clearly shows it reported a failure, but the MultiJob continued to the next entry even though that phase should have failed.
           

          John Rocha added a comment - I recently had the same problem. Where a multijob phase failed and was set to only continue if the phase was successful. However it continued even though the phase failed. This does not happen all the time. But it happens enough to cause problems. More details for our problem: Multi-job phase (unnamed) This job is set to sequential and to only continue when jobs' status are Successful sub-job#1: Run matrix job to build 32B and 64B on various machines Kill the phase on: Failure sub-job#2: Combine all of the successful build results into a build artifact and deploy to fileshare Kill the phase on Failure Multi-job phase (unnamed) Run jobs in parallel. Continue if jobs are stable or unstable but not failed Run matrix job to test 32B and 64B on various Windows OS versions Never kill this phase Multi-job phase (unnamed) Run jobs in parallel. Continue if jobs are successful Run jobs to deploy from "to-be-tested" fileshare location to production. Kill the phase on failure We had a failure during the first phase when the artifacts were being merged. The job clearly shows it reported a failure, but the MultiJob continued to the next entry even though that phase should have failed.  

            Unassigned Unassigned
            jimklimov Jim Klimov
            Votes:
            0 Vote for this issue
            Watchers:
            2 Start watching this issue

              Created:
              Updated: