[MAPREDUCE-5349] TestClusterMapReduceTestCase and TestJobName fail on Windows in branch-2

• Type: Bug
• Status: Closed
• Priority: Minor
• Resolution: Fixed
• Affects Version/s: 2.1.0-beta
• Fix Version/s: 2.1.0-beta
• Target Version/s: 2.1.0-beta
• Hadoop Flags: Reviewed
• Assignee: Chuan Liu
• Reporter: Chuan Liu
• Created: 24/Jun/13 17:29
• Updated: 27/Aug/13 22:22
• Resolved: 30/Jun/13 17:34

Description

The two unit tests fails due to MiniMRCluster use test class fullname in branch-2, instead of simple name as in trunk, to construct the MiniMRCluster identifier. Full name in the identifier almost always leads to a command script path with length larger than 260 characters which will generate an exception DefaultContainerExecutor.launchContainer() when launching the container script.

The exception looks like the follows:

2013-06-24 09:45:03,060 WARN  [ContainersLauncher #0] launcher.ContainerLaunch (ContainerLaunch.java:call(262)) - Failed to launch container.
java.io.IOException: Cannot launch container using script at path C:/Users/chuanliu/AppData/Local/Temp/1/1372092295656/org.apache.hadoop.mapred.ClusterMapReduceTestCaseConfigurableMiniMRCluster_1106798455-localDir-nm-0_1/usercache/chuanliu/appcache/application_1372092193505_0001/container_1372092193505_0001_01_000001/default_container_executor.cmd, because it exceeds the maximum supported path length of 260 characters.  Consider configuring shorter directories in yarn.nodemanager.local-dirs.
	at org.apache.hadoop.yarn.server.nodemanager.DefaultContainerExecutor.launchContainer(DefaultContainerExecutor.java:159)
	at org.apache.hadoop.yarn.server.nodemanager.containermanager.launcher.ContainerLaunch.call(ContainerLaunch.java:257)
	at org.apache.hadoop.yarn.server.nodemanager.containermanager.launcher.ContainerLaunch.call(ContainerLaunch.java:1)
	at java.util.concurrent.FutureTask$Sync.innerRun(FutureTask.java:303)
	at java.util.concurrent.FutureTask.run(FutureTask.java:138)
	at java.util.concurrent.ThreadPoolExecutor$Worker.runTask(ThreadPoolExecutor.java:886)
	at java.util.concurrent.ThreadPoolExecutor$Worker.run(ThreadPoolExecutor.java:908)
	at java.lang.Thread.run(Thread.java:662)

Issue Links

• is part of: HADOOP-9620 Umbrella jira to track windows unit test failures on branch-2
• is part of: MAPREDUCE-4401 Enhancements to MapReduce for Windows Server and Windows Azure development and runtime environments
• relates to: MAPREDUCE-5083 MiniMRCluster should use a random component when creating an actual cluster