HBASE-21421 Do not kill RS if reportOnlineRegions fails

hbase-server/src/main/java/org/apache/hadoop/hbase/master/assignment/AssignmentManager.java

@@ -1007,8 +1007,7 @@ public class AssignmentManager implements ServerListener {
     }
   }
 
-  void checkOnlineRegionsReport(final ServerStateNode serverNode, final Set<byte[]> regionNames)
-      throws YouAreDeadException {
+  void checkOnlineRegionsReport(final ServerStateNode serverNode, final Set<byte[]> regionNames) {
     final ServerName serverName = serverNode.getServerName();
     try {
       for (byte[] regionName: regionNames) {
@@ -1047,9 +1046,10 @@ public class AssignmentManager implements ServerListener {
         }
       }
     } catch (UnexpectedStateException e) {
-      LOG.warn("Killing " + serverName + ": " + e.getMessage());
-      killRegionServer(serverName);
-      throw (YouAreDeadException)new YouAreDeadException(e.getMessage()).initCause(e);
+      //See HBASE-21421, we can count on reportRegionStateTransition calls
+      //We only log a warming here. It could be a network lag.
+      LOG.warn("Failed to checkOnlineRegionsReport, maybe due to network lag, "
+          + "if this message continues, be careful of double assign", e);
     }
   }
 

hbase-server/src/test/java/org/apache/hadoop/hbase/master/assignment/TestRogueRSAssignment.java

@@ -45,6 +45,7 @@ import org.junit.AfterClass;
 import org.junit.Before;
 import org.junit.BeforeClass;
 import org.junit.ClassRule;
+import org.junit.Ignore;
 import org.junit.Rule;
 import org.junit.Test;
 import org.junit.experimental.categories.Category;
@@ -137,7 +138,11 @@ public class TestRogueRSAssignment {
     admin.setBalancerRunning(true, false);
   }
 
+  /**
+   * Ignore this test, see HBASE-21421
+   */
   @Test
+  @Ignore
   public void testReportRSWithWrongRegion() throws Exception {
     final TableName tableName = TableName.valueOf(this.name.getMethodName());