[HDFS-3933] Unclean exit in ReplicationMonitor#run occasionally causes tests to fail

• Type: Bug
• Status: Resolved
• Priority: Major
• Resolution: Duplicate
• Affects Version/s: 2.0.0-alpha
• Target Version/s: 2.0.3-alpha
• Assignee: Eli Collins
• Reporter: Eli Collins
• Created: 13/Sep/12 01:52
• Updated: 18/Sep/12 01:12
• Resolved: 18/Sep/12 01:11

Description

I've seen the following in jenkins jobs for TestHadoopArchives and TestBlockTokenWithDFS. We're not getting the initial ExitException thrown by ReplicationMonitor, just the log of the ExitException in MiniDFSCluster shutdown. Looks like the ExitException has no message so the throwable that resulted in the ExitException has no message, in which case it would be useful to capture the whole stacktrace of the throwable.

2012-09-12 18:22:08,370 FATAL [main] hdfs.MiniDFSCluster (MiniDFSCluster.java:shutdown(1355)) - Test resulted in an unexpected exit
org.apache.hadoop.util.ExitUtil$ExitException
	at org.apache.hadoop.util.ExitUtil.terminate(ExitUtil.java:86)
	at org.apache.hadoop.util.ExitUtil.terminate(ExitUtil.java:104)
	at org.apache.hadoop.hdfs.server.blockmanagement.BlockManager$ReplicationMonitor.run(BlockManager.java:3031)
	at java.lang.Thread.run(Thread.java:662)

Issue Links

• duplicates: HDFS-3936 MiniDFSCluster shutdown races with BlocksMap usage
• relates to: HADOOP-8801 ExitUtil#terminate should capture the exception stack trace