[ACCUMULO-3702] MapReduce tests fail against in LocalJobRunner with Hadoop 2.8.0-SNAPSHOT

• Type: Bug
• Status: Resolved
• Priority: Major
• Resolution: Fixed
• Fix Version/s: 1.6.3, 1.7.0
• Component/s: test
• Assignee: Josh Elser
• Reporter: Josh Elser
• Created: 31/Mar/15 01:49
• Updated: 31/Mar/15 04:00
• Resolved: 31/Mar/15 04:00

Description

Noticed that the following mapreduce tests fail against 2.8.0-SNAPSHOT. Couldn't find what changed recently.

java.io.IOException: No valid local directories in property: mapreduce.cluster.local.dir
	at org.apache.hadoop.conf.Configuration.getLocalPath(Configuration.java:2302)
	at org.apache.hadoop.mapred.JobConf.getLocalPath(JobConf.java:569)

org.apache.accumulo.core.client.mapred.AccumuloFileOutputFormatTest.testEmptyWrite
org.apache.accumulo.core.client.mapred.AccumuloFileOutputFormatTest.testRealWrite
org.apache.accumulo.core.client.mapred.AccumuloFileOutputFormatTest.writeBadVisibility
org.apache.accumulo.core.client.mapred.AccumuloInputFormatTest.testMap
org.apache.accumulo.core.client.mapred.AccumuloMultiTableInputFormatTest.testMap
org.apache.accumulo.core.client.mapred.AccumuloOutputFormatTest.testMR
org.apache.accumulo.core.client.mapred.AccumuloRowInputFormatTest.test
org.apache.accumulo.core.client.mapred.TokenFileTest.testMR
org.apache.accumulo.core.client.mapreduce.AccumuloFileOutputFormatTest.testEmptyWrite
org.apache.accumulo.core.client.mapreduce.AccumuloFileOutputFormatTest.testRealWrite
org.apache.accumulo.core.client.mapreduce.AccumuloFileOutputFormatTest.writeBadVisibility
org.apache.accumulo.core.client.mapreduce.AccumuloInputFormatTest.testMap
org.apache.accumulo.core.client.mapreduce.AccumuloInputFormatTest.testPartialFailedInputSplitDelegationToConfiguration
org.apache.accumulo.core.client.mapreduce.AccumuloInputFormatTest.testPartialInputSplitDelegationToConfiguration
org.apache.accumulo.core.client.mapreduce.AccumuloMultiTableInputFormatTest.testMap
org.apache.accumulo.core.client.mapreduce.AccumuloOutputFormatTest.testMR
org.apache.accumulo.core.client.mapreduce.AccumuloRowInputFormatTest.test
org.apache.accumulo.core.client.mapreduce.TokenFileTest.testMR
org.apache.accumulo.examples.simple.filedata.ChunkInputFormatTest.testErrorOnNextWithoutClose
org.apache.accumulo.examples.simple.filedata.ChunkInputFormatTest.test
org.apache.accumulo.examples.simple.filedata.ChunkInputFormatTest.testInfoWithoutChunks

Makes sense that we define a local.dir in the configuration. Not sure yet if it's a regression over previous Hadoop versions.