[HADOOP-1442] Zero-byte input files are not included in InputSplit

• Type: Bug
• Status: Closed
• Priority: Major
• Resolution: Fixed
• Affects Version/s: 0.12.3
• Fix Version/s: 0.14.0

Environment

ALL

• Assignee: Senthil Subramanian
• Reporter: Milind Barve
• Created: 29/May/07 23:42
• Updated: 08/Jul/09 16:51
• Resolved: 18/Jun/07 22:27

Description

The default FileInputFormat::getSplits method does not include zero-byte input files in FileSplit[]. Applications that assume the minimum number of maps to be at least number of input files can get bitten by this, as well as applications that use -reducer NONE to produce number of partitions that are equal to input partitions.