Uploaded image for project: 'OpenShift Hive'
  1. OpenShift Hive
  2. HIVE-1809

Cluster pool entry should be removed only when all clusters in the pool have been uninstalled

XMLWordPrintable

    • Icon: Story Story
    • Resolution: Duplicate
    • Icon: Normal Normal
    • None
    • None
    • False
    • None
    • False

      Today when a user deletes a clusterpool, the clusterpool CR is removed immediately after issuing deletes for all related ClusterDeployments. The clusters(clusterdeployments) are still in deprovisioning, but the ClusterPool is gone.

      Also, today claimed clusters are not affected by ClusterPool deletion. Aside from this probably being unexpected behavior, I believe it also means there's a bug: if the pool is gone, deleting a ClusterClaim on that pool will not result in the respective CD being deleted.

      AC:

      • When the user deletes a clusterpool, the clusterpool entry should be removed only when all related clusters(clusterdeployments) are finished uninstalling successfully.
      • We should honor maxConcurrent during this process. (If the user wants the deletion to happen more quickly, she would have to edit maxConcurrent before deleting the pool.)
      • The ClusterPool should not go away until claimed clusters have been dealt with. Possibilities:
        1. Consider a toggle for the user to decide whether claimed clusters should be deleted when the pool is deleted. If yes, deleting the pool causes all existing claims to be deleted, resulting in the pool deleting the CDs through the normal process.
        2. If claimed clusters are not deleted, the controller should either "orphan" those CDs (clear their ClusterPoolRef and remove their ClusterClaim so they become just like standalone CDs); or leave them alone, but keep the ClusterPool alive until they've all been deleted of their own accord (i.e. by the user deleting the ClusterClaims explicitly).

              Unassigned Unassigned
              daliu@redhat.com DangPeng Liu
              Votes:
              0 Vote for this issue
              Watchers:
              2 Start watching this issue

                Created:
                Updated:
                Resolved: