Uploaded image for project: 'JBoss Enterprise SOA Platform'
  1. JBoss Enterprise SOA Platform
  2. SOA-3118

EDS - Joining nodes ignore clustered cache

    XMLWordPrintable

    Details

    • Type: Bug
    • Status: Closed (View Workflow)
    • Priority: Major
    • Resolution: Done
    • Affects Version/s: 5.1.0 GA
    • Fix Version/s: 5.2.0 ER1
    • Component/s: EDS
    • Labels:
      None
    • Steps to Reproduce:
      Hide

      To recreate:

      1. Start 2 nodes in a cluster with a vdb deployed which has an internal materialized view (IMV).

      2. Query the IMV on node 1 to populate the cache on node 1.

      3. Query the IMV on node 2 to populate the cache on node 2 from node 1.

      4. Restart node 1 and query the IMV on node 1:
      Expectation: node 1 pulls cache from node 2
      Actual: node 1 re-queries the source

      Note: this behavior happens when the restarting in step 4 is done gracefully (and node 2 is notified node 1 is leaving the cluster) as well as when it's done with an kill -9 (simulating a failure)

      Show
      To recreate: 1. Start 2 nodes in a cluster with a vdb deployed which has an internal materialized view (IMV). 2. Query the IMV on node 1 to populate the cache on node 1. 3. Query the IMV on node 2 to populate the cache on node 2 from node 1. 4. Restart node 1 and query the IMV on node 1: Expectation: node 1 pulls cache from node 2 Actual: node 1 re-queries the source Note: this behavior happens when the restarting in step 4 is done gracefully (and node 2 is notified node 1 is leaving the cluster) as well as when it's done with an kill -9 (simulating a failure)
    • Release Notes Text:
      Hide
      https://issues.jboss.org/browse/SOA-3118

      If an internal materialized view in was being used in EDS, a node which re-joined the cluster ignored the cluster's cache. This caused synchronisation issues. A patch has resolved this issue. As a result, nodes that rejoin the cluster now know to interrogate the cache.
      Show
      https://issues.jboss.org/browse/SOA-3118 If an internal materialized view in was being used in EDS, a node which re-joined the cluster ignored the cluster's cache. This caused synchronisation issues. A patch has resolved this issue. As a result, nodes that rejoin the cluster now know to interrogate the cache.
    • Release Notes Docs Status:
      Documented as Resolved Issue

      Description

      Using an internal materialized view in EDS 5.1 with CP2, a node which re-joins the cluster is ignoring the cache available in the cluster causing out-of-sync caches.

        Gliffy Diagrams

          Attachments

            Issue Links

            is blocked by

            Quality Risk - A potential bug. TEIID-1493 In a cluster the distributed cache is cleared by a node rejoining

            • Major - A request that should be considered seriously but is not a show stopper.
            • Closed

              Activity

                People

                • Assignee:
                  van.halbert Van Halbert
                  Reporter:
                  dsteigne Debbie Steigner
                  Writer:
                  David Le Sage
                • Votes:
                  0 Vote for this issue
                  Watchers:
                  3 Start watching this issue

                  Dates

                  • Created:
                    Updated:
                    Resolved: