[HADOOP-5370] chukwa shouldn't write zero-length sink files

• Type: Improvement
• Status: Resolved
• Priority: Minor
• Resolution: Fixed
• Hadoop Flags: Reviewed
• Release Note: If there's no input in a time partition, Chukwa no longer writes zero-length sink files
• Assignee: Ariel Shemaiah Rabkin
• Reporter: Ariel Shemaiah Rabkin
• Created: 01/Mar/09 18:11
• Updated: 08/Jul/09 16:40
• Resolved: 03/Mar/09 23:19

Description

If Chukwa data arrives sporadically, then the collectors will generate a large number of zero-record .done files. These should be deleted automatically.