[TEZ-2199] updateLocalResourcesForInputSplits assumes wrongly that split data is on same FS as the default FS

• Type: Bug
• Status: Closed
• Priority: Major
• Resolution: Fixed
• Fix Version/s: 0.5.4, 0.6.1
• Assignee: Hitesh Shah
• Reporter: Hitesh Shah
• Created: 13/Mar/15 21:43
• Updated: 30/Jun/15 04:52
• Resolved: 17/Mar/15 18:04

Description

Seen in a Windows Azure scenario:

Caused by: java.io.FileNotFoundException: hdfs://namenode:9000/hive/scratch/_tez_scratch_dir/split_Map_1/job.split: No such file or directory.
at org.apache.hadoop.fs.azure.NativeAzureFileSystem.getFileStatus(NativeAzureFileSystem.java:1625)
at org.apache.tez.mapreduce.hadoop.MRInputHelpers.updateLocalResourcesForInputSplits(MRInputHelpers.java:639)
at org.apache.tez.mapreduce.hadoop.MRInputHelpers.configureMRInputWithLegacySplitGeneration(MRInputHelpers.java:115)