Apply a workaround for the isolation of quarantined node
[controller.git] / opendaylight / md-sal / sal-clustering-commons / src / main / java / org / opendaylight / controller / cluster / common / actor / QuarantinedMonitorActor.java
index 52df6ab7b388314879e72160890a624fc0104529..ff8bfc8ff62575e9e5af8271324a96501a15c90d 100644 (file)
@@ -8,10 +8,15 @@
 
 package org.opendaylight.controller.cluster.common.actor;
 
+import akka.actor.Address;
 import akka.actor.Props;
 import akka.actor.UntypedAbstractActor;
 import akka.japi.Effect;
+import akka.remote.AssociationErrorEvent;
+import akka.remote.RemotingLifecycleEvent;
 import akka.remote.ThisActorSystemQuarantinedEvent;
+import java.util.HashSet;
+import java.util.Set;
 import org.slf4j.Logger;
 import org.slf4j.LoggerFactory;
 
@@ -24,20 +29,23 @@ import org.slf4j.LoggerFactory;
  *
  */
 public class QuarantinedMonitorActor extends UntypedAbstractActor {
+    public static final String ADDRESS = "quarantined-monitor";
 
     private static final Logger LOG = LoggerFactory.getLogger(QuarantinedMonitorActor.class);
-
-    public static final String ADDRESS = "quarantined-monitor";
+    private static final Integer MESSAGE_THRESHOLD = 10;
 
     private final Effect callback;
     private boolean quarantined;
 
+    private Set<Address> addressSet = new HashSet<>();
+    private int count = 0;
+
     protected QuarantinedMonitorActor(final Effect callback) {
         this.callback = callback;
 
         LOG.debug("Created QuarantinedMonitorActor");
 
-        getContext().system().eventStream().subscribe(getSelf(), ThisActorSystemQuarantinedEvent.class);
+        getContext().system().eventStream().subscribe(getSelf(), RemotingLifecycleEvent.class);
     }
 
     @Override
@@ -62,6 +70,28 @@ public class QuarantinedMonitorActor extends UntypedAbstractActor {
 
             // execute the callback
             callback.apply();
+        } else  if (message instanceof AssociationErrorEvent) {
+            String errorMessage = message.toString();
+            LOG.trace("errorMessage:{}", errorMessage);
+            if (errorMessage.contains("The remote system has a UID that has been quarantined")) {
+                Address address = ((AssociationErrorEvent) message).getRemoteAddress();
+                addressSet.add(address);
+                count++;
+                LOG.trace("address:{} addressSet: {} count:{}", address, addressSet, count);
+                if (count >= MESSAGE_THRESHOLD && addressSet.size() > 1) {
+                    count = 0;
+                    addressSet.clear();
+                    final AssociationErrorEvent event = (AssociationErrorEvent) message;
+                    LOG.warn("Got quarantined via AssociationEvent by {}", event.remoteAddress());
+                    quarantined = true;
+
+                    // execute the callback
+                    callback.apply();
+                }
+            } else if (errorMessage.contains("The remote system explicitly disassociated")) {
+                count = 0;
+                addressSet.clear();
+            }
         }
     }