Uploaded image for project: 'Apache NiFi'
  1. Apache NiFi
  2. NIFI-5204

When node joins cluster, if a processor is stopping but cluster says the state is disabled, node ends up in inconsistent state

    Details

    • Type: Bug
    • Status: Resolved
    • Priority: Critical
    • Resolution: Fixed
    • Affects Version/s: None
    • Fix Version/s: 1.7.0
    • Component/s: None
    • Labels:
      None

      Description

      In order to make this "easy" to replicate, I did the following:

      1) Create a 2-node cluster.
      2) On both nodes, update nifi.properties to set "nifi.variable.registry.properties" to "1.properties"
      3) On both nodes, create 1.properties in $NIFI_HOME. For first node, set "sleep=2 mins" and for second node, set "sleep=0 millis"
      4) Update DebugFlow to support expression language for the "@OnStopped Pause Time"
      5) Configure flow with a DebugFlow processor. Can auto-terminate relationships and set run period to "10 secs."Set "@OnStopped Pause time" to "${sleep}"
      6) Disable DebugFlow processor.
      7) Disconnected Node 1.
      8) Go to Node 1 in browser and Start DebugFlow.
      9) Stop DebugFlow.
      10) While processor is still "stopping", go back Node 2 in browser and request that Node 1 re-join the cluster.

      Now, when Node 1 re-joins the cluster, it will attempt to disable the processor but won't be able to because the processor is still stopping. The following will be in the logs:

      2018-05-16 15:21:50,986 WARN [Reconnect to Cluster] org.apache.nifi.controller.ProcessorNode Processor cannot be disabled because its state is set to STOPPING

      So we now have a node in an inconsistent state.

      Additionally, if we now go to Node 1 in our browser and unselect all components, and attempt to STOP the process group, the request that is replicated attempts to stop the DebugFlow processor. Node 2 will now fail to stop the processor because the processor is disabled. As a result, Node 2 will now be kicked out of the cluster.

        Attachments

          Issue Links

            Activity

              People

              • Assignee:
                markap14 Mark Payne
                Reporter:
                markap14 Mark Payne
              • Votes:
                0 Vote for this issue
                Watchers:
                3 Start watching this issue

                Dates

                • Created:
                  Updated:
                  Resolved: