Uploaded image for project: 'Red Hat Data Grid'
  1. Red Hat Data Grid
  2. JDG-4092

After a "cluster shutdown" there is no way to bring nodes simple down

    XMLWordPrintable

    Details

    • Type: Bug
    • Status: New (View Workflow)
    • Priority: Critical
    • Resolution: Unresolved
    • Affects Version/s: RHDG 8.1 GA, RHDG 8.0.1 GA
    • Fix Version/s: None
    • Component/s: Clustering
    • Labels:
      None
    • Target Release:
    • Steps to Reproduce:
      Hide
      • Start a cluster with 3 nodes (one dist cache (default with file-store))
      • add some entries
      • use shutdown cluster
      • restart the complete cluster
      • shutdown each node (by CTRL-C) and wait for the StateTransfer
      • start one node only
      • try to access the cache -> IllegalArgumentException
      Show
      Start a cluster with 3 nodes (one dist cache (default with file-store)) add some entries use shutdown cluster restart the complete cluster shutdown each node (by CTRL-C) and wait for the StateTransfer start one node only try to access the cache -> IllegalArgumentException
    • Affects:
      Documentation (Ref Guide, User Guide, etc.), User Experience

      Description

      If a 3 node cluster is brough down with "shutdown cluster" the state files are created in <node>/data
      After restart it successfully and shutdown all nodes individually the expectation is that atart one node will work.
      But as the files are still there there is the
      Caused by: java.lang.IllegalArgumentException: Command does not have a topology id
      error until all nodes are back.

      Also if starting a NEW node the state is messed up as well.

      The expected behavior here, as the cluster has been scaled down to 0 (one by one) is that it must be possible to start one node as the 'new' cluster in the same way as if there was no "shutdown cluster" before.
      Note if two nodes are stopped the last node is consistent and contains all entries (as expected) and is working properly. After restarting it (which can be a long period after the others) the behavior is similar to a cluster-shutdown
      which is completely unexpected (as the other nodes might be deleted completely on purpose)

        Attachments

          Issue Links

            Activity

              People

              Assignee:
              dberinde@redhat.com Dan Berindei
              Reporter:
              rhn-support-wfink Wolf-Dieter Fink
              Votes:
              0 Vote for this issue
              Watchers:
              1 Start watching this issue

                Dates

                Created:
                Updated: