Uploaded image for project: 'OpenFlowPlugin'
  1. OpenFlowPlugin
  2. OPNFLWPLUG-1086

Reconciliation framework failure when starting cbench tool for the first time

    XMLWordPrintable

Details

    Description

      this issue was noticed because an old CSIT job using cbench had a consistent failure that is not
      seen in sodium (technically a regression in that sense)

      The cbench tool fails on starting with a "closed connection ... exiting" message.

      snippet from this karaf.log :

      2020-02-22T03:05:59,721 | INFO  | opendaylight-cluster-data-akka.actor.default-dispatcher-6 | GuardedContextImpl               | 310 - org.opendaylight.openflowplugin.impl - 0.10.0.SNAPSHOT | Starting RpcContextImpl[NEW] service for node openflow:0
      2020-02-22T03:05:59,721 | WARN  | ofppool-7        | ContextChainHolderImpl           | 310 - org.opendaylight.openflowplugin.impl - 0.10.0.SNAPSHOT | Reconciliation framework failure.
      2020-02-22T03:05:59,722 | INFO  | ofppool-7        | ReconciliationManagerImpl        | 302 - org.opendaylight.openflowplugin.applications.reconciliation-framework - 0.10.0.SNAPSHOT | Stopping reconciliation for node Uri{_value=openflow:10}
      2020-02-22T03:05:59,726 | WARN  | ofppool-0        | ContextChainHolderImpl           | 310 - org.opendaylight.openflowplugin.impl - 0.10.0.SNAPSHOT | Reconciliation framework failure.
      2020-02-22T03:05:59,726 | INFO  | ofppool-0        | ReconciliationManagerImpl        | 302 - org.opendaylight.openflowplugin.applications.reconciliation-framework - 0.10.0.SNAPSHOT | Stopping reconciliation for node Uri{_value=openflow:5}
      2020-02-22T03:05:59,738 | INFO  | ListenableFutureAdapter-thread-0 | ContextChainHolderImpl           | 310 - org.opendaylight.openflowplugin.impl - 0.10.0.SNAPSHOT | Device openflow:3 connection is enabled by reconciliation framework.
      

      Only the first try fails though, and additional cbench tests are ok.

      We should figure out the root cause of the failure before we talk about the few workarounds
      I know we can do:

      1. start a dummy cbench test first to get past the initial failure
      2. remove the cbench tests altogether if they no longer provide any value, although I do check from time to time the graphs that those jobs produce

      Attachments

        Issue Links

          # Subject Branch Project Status CR V

          Activity

            People

              gobinath Gobinath Suganthan
              jluhrsen Jamo Luhrsen
              Votes:
              0 Vote for this issue
              Watchers:
              5 Start watching this issue

              Dates

                Created:
                Updated:
                Resolved: