The patch for trunk

The earlier patch went out of sync. Updated patch.

Please add a call to UserGroupInformation.setConfiguration at the start of the JobTracker, and TaskTracker so that UGI uses the same configuration object. The Task should also use setConfiguration with the job's configuration.

Please use the long form of the user name rather than the short form in spots like:

-    JobInfo jobInfo = new JobInfo(jobId, new Text(ugi.getUserName()), 
+    JobInfo jobInfo = new JobInfo(jobId, new Text(ugi.getShortUserName()), 

The following code from Cluster.java and ClusterWithLinuxTaskController is a no-op:

UserGroupInformation.getCurrentUser().doAs(...)

Patch that addresses Owen's concerns. Running tests now. Running test-patch proved to be a real pain. Still trying to run it..

Tests pass with this patch.

Ok i could run findbugs on this patch and confirmed that it doesn't add any new warnings. I also ran javadoc and javac-warnings ones. No new warnings there too. I am sure I didn't introduce any new release audit warnings since there are no new files in the patch. Same for @author tags. It is almost impossible to run the stand-alone test-patch command, as it exists today, on this patch.

+1

I just committed this. Thanks, Devaraj!

Integrated in Hadoop-Mapreduce-trunk #221 (See http://hudson.zones.apache.org/hudson/job/Hadoop-Mapreduce-trunk/221/)

Integrated in Hadoop-Mapreduce-trunk-Commit #225 (See http://hudson.zones.apache.org/hudson/job/Hadoop-Mapreduce-trunk-Commit/225/)

Integrated in Hadoop-Mapreduce-trunk-Commit #255 (See http://hudson.zones.apache.org/hudson/job/Hadoop-Mapreduce-trunk-Commit/255/)
MAPREDUCE-1421. LinuxTaskController tests failing on trunk after the commit of . Contributed by Amareshwari Sriramadasu.

Integrated in Hadoop-Mapreduce-trunk #248 (See http://hudson.zones.apache.org/hudson/job/Hadoop-Mapreduce-trunk/248/)