hadoop-common-dev mailing list archives

Site index · List index
Message view « Date » · « Thread »
Top « Date » · « Thread »
From "Jim Kellerman (JIRA)" <j...@apache.org>
Subject [jira] Commented: (HADOOP-2323) JobTracker.close() prints stack traces for exceptions that are not errors
Date Mon, 03 Dec 2007 01:46:52 GMT

    [ https://issues.apache.org/jira/browse/HADOOP-2323?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel#action_12547711

Jim Kellerman commented on HADOOP-2323:


I agree with you, but not knowing the internals of the JobTracker that well, did not know
if there was a reason that JobTracker$ExpireTrackers and JobTracker$TaskCommitQueue were different
from the other threads and intentionally let the exception escape rather than shut down.

Most of the other threads also use a class-level boolean but that is probably not necessary.
A boolean in the method (or just breaking out of the loop) would work just as well.

I'll work up a more minimal patch along these lines and you (or someone else who knows JobTracker)
can decide which patch you like.


> JobTracker.close() prints stack traces for exceptions that are not errors
> -------------------------------------------------------------------------
>                 Key: HADOOP-2323
>                 URL: https://issues.apache.org/jira/browse/HADOOP-2323
>             Project: Hadoop
>          Issue Type: Bug
>          Components: mapred
>    Affects Versions: 0.16.0
>            Reporter: Jim Kellerman
>            Assignee: Jim Kellerman
>             Fix For: 0.16.0
>         Attachments: patch.txt
> JobTracker.close() prints a stack trace for an interrupted exception even though it was
the method that interrupted the thread that threw the exception. For example:
> {code}
>       this.expireTrackers.stopTracker();
>       try {
>         this.expireTrackersThread.interrupt();
>         this.expireTrackersThread.join();
>       } catch (InterruptedException ex) {
>         ex.printStackTrace();
>       }
> {code}
> Well of course it is going to catch an InterruptedException after it just interrupted
the thread!
> This is *not* an error and should  *not* be dumped to the logs!
> In other circumstances, catching InterruptedException is entirely appropriate. Just not
in close where you've told the thread to shutdown and then interrupted it to ensure it does!

This message is automatically generated by JIRA.
You can reply to this email to add a comment to the issue online.

View raw message