[HADOOP-5400] JT restart recovery: Exclude jobs which failed during SUBMIT_JOB (due to acl)

• Type: Bug
• Status: Closed
• Priority: Blocker
• Resolution: Duplicate

Environment

Hadoop 0.20 + 0.20.0 + HADOOP-5225 + HADOOP-5332

• Assignee: Unassigned
• Reporter: Rajiv Chittajallu
• Created: 04/Mar/09 20:29
• Updated: 08/Jul/09 16:53
• Resolved: 05/Mar/09 04:58

Description

mapred.jobtracker.restart.recover is set to true in mapred-site.xml

This is a job that failed during Job submit due to invalid ACL

2009-03-04 18:31:25,970 INFO org.apache.hadoop.ipc.Server: IPC Server handler 14 on 50300, call submitJob(job_200903041223_0259) from 192.168.10.1:41306: error: org.apache.hadoop.security.AccessControlException: User rajive cannot perform operation SUBMIT_JOB on queue default

When the JobTracker was restarted after some time, the failed job was being recovered/restarted

2009-03-04 19:13:30,544 INFO org.apache.hadoop.mapred.JobTracker: Found an incomplete job directory job_200903041852_0040. Deleting it!!
2009-03-04 19:13:30,613 INFO org.apache.hadoop.mapred.FairScheduler: Successfully configured FairScheduler
2009-03-04 19:13:30,614 INFO org.apache.hadoop.mapred.JobTracker: Trying to recover job job_200903041223_0259

2009-03-04 18:53:17,147 INFO org.apache.hadoop.mapred.JobTracker: JobTracker failed to recover job job_200903041223_0259. Ignoring it.
java.io.FileNotFoundException: File file:/var/log/hadoop//history/jobtracker1.foo.com_1236192735577_job_200903041223_0259_rajive_word+count does not exist.
at org.apache.hadoop.fs.RawLocalFileSystem.getFileStatus(RawLocalFileSystem.java:360)
at org.apache.hadoop.fs.FilterFileSystem.getFileStatus(FilterFileSystem.java:245)
at org.apache.hadoop.fs.ChecksumFileSystem$ChecksumFSInputChecker.<init>(ChecksumFileSystem.java:125)
at org.apache.hadoop.fs.ChecksumFileSystem.open(ChecksumFileSystem.java:283)
at org.apache.hadoop.fs.FileSystem.open(FileSystem.java:336)
at org.apache.hadoop.mapred.JobHistory.parseHistoryFromFS(JobHistory.java:245)
at org.apache.hadoop.mapred.JobTracker$RecoveryManager.recover(JobTracker.java:1144)
at org.apache.hadoop.mapred.JobTracker.offerService(JobTracker.java:1603)
at org.apache.hadoop.mapred.JobTracker.main(JobTracker.java:3326)
2009-03-04 18:53:17,147 INFO org.apache.hadoop.mapred.JobTracker: Restart count for job job_200903041223_0259 is 0
2009-03-04 18:53:18,626 INFO org.apache.hadoop.mapred.JobInProgress: Input size for job job_200903041223_0259 = 4664646202464
2009-03-04 18:53:18,626 INFO org.apache.hadoop.mapred.JobInProgress: Split info for job:job_200903041223_0259 with 34640 splits:

These jobs failed during job submit shouldn't be considered for recovery.