Uploaded image for project: 'OpenShift Bugs'
  1. OpenShift Bugs
  2. OCPBUGS-13107

cluster-destroy: Too many workers for a standalone OSP

    XMLWordPrintable

Details

    • -
    • Low
    • No
    • ShiftStack Sprint 235
    • 1
    • False
    • Hide

      None

      Show
      None
    • None

    Description

      Description of problem:

      Running `openshift-install cluster destroy` defeats an OpenStack cloud with many Swift objects, if said cloud is low on resources.
      
      In particular, testing the teardown of an OCP cluster with 500.000 objects in the image registry caused RabbitMQ to crash on a standalone (single-host) OpenStack deployment backed with NVMe storage.
      

      Version-Release number of selected component (if applicable):

      
      

      How reproducible:

      on a constrained (single-host) OpenStack cloud, with the default limit of 10000 to the bulk-deletion of Swift objects.
      

      Steps to Reproduce:

      1. install OpenShift
      2. upload 500000 arbitrary objects in the image-registry container
      3. launch cluster teardown
      4. enjoy Swift responding 504 errors, and the rest of the cluster to become unstable
      

      Attachments

        Activity

          People

            pprinett@redhat.com Pierre Prinetti
            pprinett@redhat.com Pierre Prinetti
            Itshak Brown Itshak Brown
            Votes:
            0 Vote for this issue
            Watchers:
            4 Start watching this issue

            Dates

              Created:
              Updated:
              Resolved: