[MAPREDUCE-6762] ControlledJob#toString failed with NPE when job status is not successfully updated

• Type: Bug
• Status: Resolved
• Priority: Major
• Resolution: Fixed
• Affects Version/s: 2.7.2
• Fix Version/s: 2.9.0, 3.0.0-alpha1
• Hadoop Flags: Reviewed
• Assignee: Weiwei Yang
• Reporter: Weiwei Yang
• Created: 19/Aug/16 04:07
• Updated: 31/Jan/19 13:19
• Resolved: 21/Aug/16 16:22

Description

This issue was found from a cluster where Pig query occasionally failed on NPE. Pig uses JobControl API to track MR job status, but sometimes Job History Server failed to flush job meta files to HDFS which caused the status update failed. Then we get NPE in org.apache.hadoop.mapreduce.Job.getJobName. The result of this situation is quite confusing: Pig query failed, job history is missing, but the job status on Yarn is succeed.

Issue Links

• is related to: PIG-4967 NPE in PigJobControl.run() when job status is null
• is related to: MAPREDUCE-7179 Job status not available when job finished