[
https://issues.apache.org/jira/browse/HADOOP-1350?page=com.atlassian.jira.plugin.system.issuetabpanels:all-tabpanel
]
Devaraj Das updated HADOOP-1350:
--------------------------------
Attachment: 1350.patch
Changed the chunked-encoding handling. Now the servlet doesn't set the header
for chunked-encoding explicitly. Instead it leaves it to jetty to do
chunked-encoding if required. Looks like there is an issue if we set that
header explicitly.
sort500 works fine with this patch. I am yet to test the huge map output
problem (the original problem reported in HADOOP-1176).
> Shuffle started taking a very long time after the HADOOP-1176 fix
> -----------------------------------------------------------------
>
> Key: HADOOP-1350
> URL: https://issues.apache.org/jira/browse/HADOOP-1350
> Project: Hadoop
> Issue Type: Bug
> Components: mapred
> Affects Versions: 0.13.0
> Reporter: Devaraj Das
> Assigned To: Devaraj Das
> Priority: Blocker
> Fix For: 0.13.0
>
> Attachments: 1350.patch
>
>
> Shuffle phase started taking a very long time after the fix for HADOOP-1176.
--
This message is automatically generated by JIRA.
-
You can reply to this email to add a comment to the issue online.
