Uploaded image for project: 'Data Management'
  1. Data Management
  2. DM-34740

Jenkins script checking sciplat-lab build completion exits prematurely

    XMLWordPrintable

    Details

    • Type: Bug
    • Status: Done
    • Resolution: Done
    • Fix Version/s: None
    • Component/s: jenkins
    • Labels:
      None
    • Story Points:
      2
    • Team:
      SQuaRE
    • Urgent?:
      No

      Description

      Today, 2022-05-10, the sqre/infra/build-sciplatlab job exited with success status after just 36 seconds. I believe this happened because the GitHub Action took 2 minutes to start (triggered at 1311Z and started at 1313Z). The pipeline script waits only 30 seconds for it to start. As a result, the status from the previous execution was used.

      A simple fix would be to check that the created_at time in the job status result is recent, waiting until it is not more than a minute or so old. (Of course, subsequent status checks while the job is running should not check this.)

        Attachments

          Activity

          Hide
          athornton Adam Thornton added a comment -

          So after much hilarity...

           

          Now we're detecting the start time, but we get confused if anyone else starts a new build job while we're polling.  I have an idea.

          Show
          athornton Adam Thornton added a comment - So after much hilarity...   Now we're detecting the start time, but we get confused if anyone else starts a new build job while we're polling.  I have an idea.
          Hide
          athornton Adam Thornton added a comment -

          Poll until we capture a job ID (the ID of the first job started later than a second before the integer-number-of-seconds timestamp before the job submission, so that could be the wrong job if we're very unlucky), and then track that job to completion.

          Done, reviewed, merged, tested.

          Show
          athornton Adam Thornton added a comment - Poll until we capture a job ID (the ID of the first job started later than a second before the integer-number-of-seconds timestamp before the job submission, so that could be the wrong job if we're very unlucky), and then track that job to completion. Done, reviewed, merged, tested.

            People

            Assignee:
            athornton Adam Thornton
            Reporter:
            ktl Kian-Tat Lim
            Watchers:
            Adam Thornton, Kian-Tat Lim
            Votes:
            0 Vote for this issue
            Watchers:
            2 Start watching this issue

              Dates

              Created:
              Updated:
              Resolved:

                Jenkins

                No builds found.