Apache
/
hadoop
1
0
Fork 0
Code Issues Pull Requests Packages Projects Releases Wiki Activity

MAPREDUCE-3837. Job tracker is not able to recover jobs after crash. Contributed by Mayank Bansal. 

git-svn-id: https://svn.apache.org/repos/asf/hadoop/common/branches/branch-0.23@1243698 13f79535-47bb-0310-9956-ffa450edef68
This commit is contained in:
Konstantin Shvachko 2012-02-13 21:16:22 +00:00
parent 87cd58b0f9
commit 10c7f0eb15
2 changed files with 12 additions and 5 deletions
Show Stats Download Patch File Download Diff File Expand all files Collapse all files

3
hadoop-mapreduce-project/CHANGES.txt
View File

@ -2649,6 +2649,9 @@ Release 0.22.1 - Unreleased
BUG FIXES BUG FIXES
MAPREDUCE-3837. Job tracker is not able to recover jobs after crash.
(Mayank Bansal via shv)
Release 0.22.0 - 2011-11-29 Release 0.22.0 - 2011-11-29
INCOMPATIBLE CHANGES INCOMPATIBLE CHANGES

14
hadoop-mapreduce-project/src/java/org/apache/hadoop/mapred/JobTracker.java
View File

@ -1192,13 +1192,17 @@ public class JobTracker implements MRConstants, InterTrackerProtocol,
try { try {
Path jobInfoFile = getSystemFileForJob(jobId); Path jobInfoFile = getSystemFileForJob(jobId);
FSDataInputStream in = fs.open(jobInfoFile); FSDataInputStream in = fs.open(jobInfoFile);
JobInfo token = new JobInfo(); final JobInfo token = new JobInfo();
token.readFields(in); token.readFields(in);
in.close(); in.close();
UserGroupInformation ugi = final UserGroupInformation ugi =
UserGroupInformation.createRemoteUser(token.getUser().toString()); UserGroupInformation.createRemoteUser(token.getUser().toString());
submitJob(token.getJobID(), restartCount, ugi.doAs(new PrivilegedExceptionAction<JobStatus>() {
ugi, token.getJobSubmitDir().toString(), true, null); public JobStatus run() throws IOException ,InterruptedException{
return submitJob(token.getJobID(), restartCount,
ugi, token.getJobSubmitDir().toString(), true, null);
}});
recovered++; recovered++;
} catch (Exception e) { } catch (Exception e) {
LOG.warn("Could not recover job " + jobId, e); LOG.warn("Could not recover job " + jobId, e);