[HDFS-13799] TestEditLogTailer#testTriggersLogRollsForAllStandbyNN fails due to missing synchronization between rollEditsRpcExecutor and tailerThread shutdown

• Type: Bug
• Status: Resolved
• Priority: Minor
• Resolution: Fixed
• Affects Version/s: 3.0.0
• Fix Version/s: 3.2.0, 3.1.2
• Component/s: ha
• Hadoop Flags: Reviewed
• Assignee: Hrishikesh Gadre
• Reporter: Hrishikesh Gadre
• Created: 06/Aug/18 21:04
• Updated: 04/Apr/19 16:47
• Resolved: 07/Aug/18 23:14

Description

TestEditLogTailer#testTriggersLogRollsForAllStandbyNN unit test is failing in our internal environment with following error,

java.lang.AssertionError: Test resulted in an unexpected exit
	at org.apache.hadoop.hdfs.server.namenode.ha.TestEditLogTailer.testTriggersLogRollsForAllStandbyNN(TestEditLogTailer.java:245)

This test failure is due to following error during shutdown of the MiniDfsCluster

2018-07-31 21:59:27,806 [main] INFO  hdfs.MiniDFSCluster (MiniDFSCluster.java:shutdown(1965)) - Shutting down the Mini HDFS Cluster
2018-07-31 21:59:27,806 [main] FATAL hdfs.MiniDFSCluster (MiniDFSCluster.java:shutdown(1968)) - Test resulted in an unexpected exit
1: java.util.concurrent.RejectedExecutionException: Task java.util.concurrent.FutureTask@1ce1d2b6 rejected from java.util.concurrent.ThreadPoolExecutor@12263f5a[Terminated, pool size = 0, active threads = 0, queued tasks = 0, completed tasks = 0]
	at org.apache.hadoop.util.ExitUtil.terminate(ExitUtil.java:265)
	at org.apache.hadoop.hdfs.server.namenode.ha.EditLogTailer$EditLogTailerThread.doWork(EditLogTailer.java:441)
	at org.apache.hadoop.hdfs.server.namenode.ha.EditLogTailer$EditLogTailerThread.access$400(EditLogTailer.java:380)
	at org.apache.hadoop.hdfs.server.namenode.ha.EditLogTailer$EditLogTailerThread$1.run(EditLogTailer.java:397)
	at org.apache.hadoop.security.SecurityUtil.doAsLoginUserOrFatal(SecurityUtil.java:482)
	at org.apache.hadoop.hdfs.server.namenode.ha.EditLogTailer$EditLogTailerThread.run(EditLogTailer.java:393)
Caused by: java.util.concurrent.RejectedExecutionException: Task java.util.concurrent.FutureTask@1ce1d2b6 rejected from java.util.concurrent.ThreadPoolExecutor@12263f5a[Terminated, pool size = 0, active threads = 0, queued tasks = 0, completed tasks = 0]
	at java.util.concurrent.ThreadPoolExecutor$AbortPolicy.rejectedExecution(ThreadPoolExecutor.java:2047)
	at java.util.concurrent.ThreadPoolExecutor.reject(ThreadPoolExecutor.java:823)
	at java.util.concurrent.ThreadPoolExecutor.execute(ThreadPoolExecutor.java:1369)
	at java.util.concurrent.AbstractExecutorService.submit(AbstractExecutorService.java:134)
	at java.util.concurrent.Executors$DelegatedExecutorService.submit(Executors.java:681)
	at org.apache.hadoop.hdfs.server.namenode.ha.EditLogTailer.triggerActiveLogRoll(EditLogTailer.java:351)
	at org.apache.hadoop.hdfs.server.namenode.ha.EditLogTailer$EditLogTailerThread.doWork(EditLogTailer.java:411)
	... 4 more

It looks like the EditLogTailer class is not handling the shutdown correctly. Specifically the EditLogTailer#stop() method shuts down the rollEditsRpcExecutor executor service before setting the tailerThread#shouldRun flag. This is a race condition since the tailerThread can try to submit a new task to this executor service which has been asked to shutdown. If that happens, it will receive an unexpected RejectedExecutionException, resulting in a test failure. The solution should be to properly synchronize shutdown of tailerThread with rollEditsRpcExecutor.

Issue Links

• relates to: HDFS-14415 Backport HDFS-13799 to branch-2