[MAPREDUCE-3933] Failures because MALLOC_ARENA_MAX is not set

• Type: Bug
• Status: Closed
• Priority: Major
• Resolution: Fixed
• Affects Version/s: 0.23.0
• Fix Version/s: 2.0.0-alpha
• Component/s: mrv2, test
• Hadoop Flags: Reviewed
• Assignee: Ahmed Radwan
• Reporter: Ahmed Radwan
• Created: 28/Feb/12 04:57
• Updated: 23/May/12 20:28
• Resolved: 29/Feb/12 16:22

Description

We have noticed a bunch of MapReduce test failures on CentOS 6 due to "running beyond virtual memory limits".

These tests fail with messages of the form:

[Node Status Updater] nodemanager.NodeStatusUpdaterImpl (NodeStatusUpdaterImpl.java:getNodeStatus(254)) - Sending out status for container: container_id {, app_attempt_id {, application_id {, id: 1, cluster_timestamp: 1330401645767, }, attemptId: 1, }, id: 1, }, state: C_RUNNING, diagnostics: "Container [pid=16750,containerID=container_1330401645767_0001_01_000001] is running beyond virtual memory limits. Current usage: 220.5mb of 2.0gb physical memory used; 7.1gb of 4.2gb virtual memory used. Killing container

The failing tests are:

TestJobCounters
TestJobSysDirWithDFS
TestLazyOutput
TestMiniMRChildTask
TestMiniMRClientCluster
TestReduceFetchFromPartialMem
TestChild
TestMapReduceLazyOutput
TestJobOutputCommitter
TestMRAppWithCombiner
TestMRJobs
TestMRJobsWithHistoryService
TestMROldApiJobs
TestSpeculativeExecution
TestUberAM

I'll upload a patch momentarily.

Issue Links

• relates to: HBASE-5529 MR test failures becuase MALLOC_ARENA_MAX is not set