[HIVE-1403] Reporting progress to JT during closing files in FileSinkOperator

• Type: Bug
• Status: Closed
• Priority: Major
• Resolution: Fixed
• Fix Version/s: 0.6.0
• Hadoop Flags: Reviewed
• Assignee: Ning Zhang
• Reporter: Ning Zhang
• Created: 14/Jun/10 20:15
• Updated: 17/Dec/11 00:03
• Resolved: 15/Jun/10 15:31

Description

If there are too many files need to be closed in FileSinkOperator (e.g., if DynamicPartition/FileSpray is turned on), there could be many files generated by each task and they need to be closed at the FileSinkOperator.closeOp(). If the NN is overloaded each file close could take more than 1 sec. This sometimes make JT think the task is dead since it takes too long to close all the files and without any progress report. We need to report progress after a while during file closing.