Guanghao Zhang created HBASE-24897: -------------------------------------- Summary: Fix flaky test TestRegionReplicaReplicationEndpoint Key: HBASE-24897 URL: https://issues.apache.org/jira/browse/HBASE-24897 Project: HBase Issue Type: Bug Reporter: Guanghao Zhang
Debug this unti test, I found the RS aborted because RegionReplicaFlushHandler flush failed. When create a new table with region replica, the assign order may be: # assign 0002 replica region and trigger primary region flush. # assign 0001 replica region and trigger primary region flush. # assign primary region. But the primary region flush may failed because the primary region not opened now. So it may abort the RS...... {code:java} 2020-08-18 16:56:30,041 INFO [RS_OPEN_REGION-regionserver/hao-OptiPlex-7050:0-0] handler.AssignRegionHandler(141): Opened testRegionReplicaReplicationIgnoresDisabledTables_drop_false_disabledReplication_false,,1597740978463_0002.66e9757a05fbae7623cfea3369fc8354. 2020-08-18 16:56:30,558 INFO [RS_OPEN_REGION-regionserver/hao-OptiPlex-7050:0-0] handler.AssignRegionHandler(141): Opened testRegionReplicaReplicationIgnoresDisabledTables_drop_false_disabledReplication_false,,1597740978463_0001.22ff45423b0f1f0e93794f673449d140. 2020-08-18 16:56:31,192 INFO [RS_OPEN_REGION-regionserver/hao-OptiPlex-7050:0-0] handler.AssignRegionHandler(141): Opened testRegionReplicaReplicationIgnoresDisabledTables_drop_false_disabledReplication_false,,1597740978463.901f9cd06bbf27ef7c2d70b5af725cd2. 2020-08-18 16:58:53,857 ERROR [RS_REGION_REPLICA_FLUSH_OPS-regionserver/hao-OptiPlex-7050:0-0] helpers.MarkerIgnoringBase(159): ***** ABORTING region server hao-optiplex-7050,36368,1597740961432: ServerAborting because an exception was thrown ***** org.apache.hadoop.hbase.client.NoServerForRegionException: No server address listed in hbase:meta for region testRegionReplicaReplicationWithReplicas_10,,1597741128945.0f541dc1a7ca64797c4cf054adb9edfb. containing row at org.apache.hadoop.hbase.client.ConnectionImplementation.locateRegionInMeta(ConnectionImplementation.java:926) at org.apache.hadoop.hbase.client.ConnectionImplementation.locateRegion(ConnectionImplementation.java:784) at org.apache.hadoop.hbase.client.ConnectionUtils$ShortCircuitingClusterConnection.locateRegion(ConnectionUtils.java:140) at org.apache.hadoop.hbase.client.RegionAdminServiceCallable.getRegionLocations(RegionAdminServiceCallable.java:147) at org.apache.hadoop.hbase.client.RegionAdminServiceCallable.getLocation(RegionAdminServiceCallable.java:98) at org.apache.hadoop.hbase.client.RegionAdminServiceCallable.prepare(RegionAdminServiceCallable.java:84) at org.apache.hadoop.hbase.client.FlushRegionCallable.prepare(FlushRegionCallable.java:62) at org.apache.hadoop.hbase.client.RpcRetryingCallerImpl.callWithRetries(RpcRetryingCallerImpl.java:105) at org.apache.hadoop.hbase.regionserver.handler.RegionReplicaFlushHandler.triggerFlushInPrimaryRegion(RegionReplicaFlushHandler.java:129) at org.apache.hadoop.hbase.regionserver.handler.RegionReplicaFlushHandler.process(RegionReplicaFlushHandler.java:78) at org.apache.hadoop.hbase.executor.EventHandler.run(EventHandler.java:104) at java.util.concurrent.ThreadPoolExecutor.runWorker(ThreadPoolExecutor.java:1149) at java.util.concurrent.ThreadPoolExecutor$Worker.run(ThreadPoolExecutor.java:624) at java.lang.Thread.run(Thread.java:748) {code} I thought the fix should be assign primary region firstly when enable region replica featue. Will check the implmenation of region replica. -- This message was sent by Atlassian Jira (v8.3.4#803005)