[
https://issues.apache.org/jira/browse/MAPREDUCE-3896?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=13213914#comment-13213914
]
John George commented on MAPREDUCE-3896:
----------------------------------------
Pig job through oozie hangs in certain cases due to a race condition. The exact
scenario is as follows:
Oozie launches a job which runs a mapper to launch a pig action. In certain
cases, it can be seen that the AM (pig launcher) that started the mapper task
is stuck waiting on getCounters(). The pig action itself had completed
successfully. The mapper task that was stuck seemed to be talking to the JHS to
get counters. JHS itself was throwing an AccessControlException.
{quote}
012-02-21 23:59:59,997 INFO org.apache.hadoop.ipc.Server: IPC Server listener
on 10020: readAndProcess threw exception
org.apache.hadoop.security.AccessControlException: Can't retrieve username from
tokenIdentifier. from client 98.137.97.247. Count of bytes read: 0
[Socket Reader #5 for port
10020]org.apache.hadoop.security.AccessControlException: Can't retrieve
username from tokenIdentifier.
at
org.apache.hadoop.ipc.Server$Connection.getAuthorizedUgi(Server.java:997)
at
org.apache.hadoop.ipc.Server$Connection.saslReadAndProcess(Server.java:1090)
at
org.apache.hadoop.ipc.Server$Connection.readAndProcess(Server.java:1235)
at org.apache.hadoop.ipc.Server$Listener.doRead(Server.java:570)
at
org.apache.hadoop.ipc.Server$Listener$Reader.doRunLoop(Server.java:369)
at org.apache.hadoop.ipc.Server$Listener$Reader.run(Server.java:344)
{quote}
It seems that if the mapper task calls getCounter before the pig action AM
dies, this oozie job succeeds, otherwise fails..
> pig job through oozie hangs
> ----------------------------
>
> Key: MAPREDUCE-3896
> URL: https://issues.apache.org/jira/browse/MAPREDUCE-3896
> Project: Hadoop Map/Reduce
> Issue Type: Bug
> Affects Versions: 0.23.1, 0.24.0, 0.23.2
> Reporter: John George
> Assignee: Mahadev konar
> Priority: Blocker
>
> running pig job on oozie hangs due to race condition
--
This message is automatically generated by JIRA.
If you think it was sent incorrectly, please contact your JIRA administrators:
https://issues.apache.org/jira/secure/ContactAdministrators!default.jspa
For more information on JIRA, see: http://www.atlassian.com/software/jira
