Re: Randomly failing test

2012-06-07 Thread Chris Douglas
hat is possible(either tuning some of > the parameters or increasing child JVM memory?) > > Thanks > -- Asokan > > From: Chris Douglas [cdoug...@apache.org] > Sent: Thursday, June 07, 2012 9:02 PM > To: common-dev@hadoop.apache.org >

RE: Randomly failing test

2012-06-07 Thread Asokan, M
(either tuning some of the parameters or increasing child JVM memory?) Thanks -- Asokan From: Chris Douglas [cdoug...@apache.org] Sent: Thursday, June 07, 2012 9:02 PM To: common-dev@hadoop.apache.org Subject: Re: Randomly failing test The output on MAPREDUCE

Re: Randomly failing test

2012-06-07 Thread Chris Douglas
The output on MAPREDUCE-4094 is no longer available, but a failed job during an integration test could be traced to nondeterminism anywhere in that chain. The "flakiness" you observe is common to many Hadoop integration tests. The purpose of the test is to verify that 1) the merge works when some

Randomly failing test

2012-06-07 Thread Asokan, M
Can someone guide me in making the test org.apache.hadoop.mapred.TestReduceFetchFromPartialMem less flaky? I see test failures mentioned in MAPREDUCE-4094 and MAPREDUCE-4096 but no fix is available so far. Thanks. -- Asokan ATTENTION: - The informatio