[HBASE-20678] NPE in ReplicationSourceManager#NodeFailoverWorker

• Type: Bug
• Status: Resolved
• Priority: Minor
• Resolution: Fixed
• Fix Version/s: 2.1.0
• Assignee: Guanghao Zhang
• Reporter: Guanghao Zhang
• Created: 04/Jun/18 03:06
• Updated: 01/Aug/18 06:24
• Resolved: 05/Jun/18 08:16

Description

2018-06-04 10:28:43,362 INFO [ReplicationExecutor-0] replication.ZKReplicationQueueStorage(432): Claim queue queueId=1 from hao-optiplex-7050,38491,1528079278158 to hao-optiplex-7050,39931,1528079278272 failed with org.apache.zookeeper.KeeperException$NoNodeException: KeeperErrorCode = NoNode, someone else took the log?
Exception in thread "ReplicationExecutor-0" java.lang.NullPointerException
at org.apache.hadoop.hbase.replication.regionserver.ReplicationSourceManager$NodeFailoverWorker.run(ReplicationSourceManager.java:858)
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)

ZKReplicationQueueStorage's claimQueue method may return null when got NoNodeException.

          Pair<String, SortedSet<String>> peer = queueStorage.claimQueue(deadRS,
            queues.get(ThreadLocalRandom.current().nextInt(queues.size())), server.getServerName());
          long sleep = sleepBeforeFailover / 2;
          if (!peer.getSecond().isEmpty()) {
            newQueues.put(peer.getFirst(), peer.getSecond());
            sleep = sleepBeforeFailover;
          }