Uploaded image for project: 'CloverDX'
  1. CloverDX
  2. CLO-14472

Prevent duplication of failing executions in cluster

    XMLWordPrintable

    Details

    • Type: Story
    • Status: Closed
    • Priority: Major
    • Resolution: Fixed
    • Affects Version/s: None
    • Fix Version/s: rel-5-1-0
    • Component/s: None
    • Security Level: Users (General product issues)

      Description

      Summary

      In cluster, an execution triggered by Scheduler or Event Listener can be duplicated as many times as there are nodes in the cluster. Duplication happens only if the executed job fails. For example simple click on "Run now" of Scheduler can trigger 8 job executions on 8-node cluster. This is potentially dangerous behavior and doesn't fit with other server features - e.g. jobflow failure listener will react 8 times (e.g. send 8 emails to support) instead of once.

      Details:

      Processing of a server event (Scheduler, Event Listener, ...) in cluster tries to make the triggered task successful using all allowed cluster nodes. When the task is job execution and the job fails, cluster tries to execute the job again on a different node. That means for failing job the execution is repeated multiple times per one event.

       

      Note

      Just changing this behaviour would cause backward compatibility (which might be OK if we decide so), but there still should be a way, how to let schedules/listeners operate in similar mode as they do now. User should be able to decide if he wants reprocessing on another node when the job fails.

      Acceptance criteria

      • It is possible to specify if schedule/listener should restart job on different node in case of failure
      • This option will be available only for cluster environment
      • Default for this options is ?TBD?
      • Jobs reprocessing works as configured (use other nodes only if configured)

        Attachments

          Issue Links

            Activity

              People

              • Assignee:
                sedlacekj Jan Sedlacek
                Reporter:
                salamonp Pavel Salamon
              • Votes:
                0 Vote for this issue
                Watchers:
                4 Start watching this issue

                Dates

                • Created:
                  Updated:
                  Resolved: