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)
