hadoop-common-dev mailing list archives

Site index · List index
Message view « Date » · « Thread »
Top « Date » · « Thread »
From "Arun C Murthy (JIRA)" <j...@apache.org>
Subject [jira] Commented: (HADOOP-3280) virtual address space limits break streaming apps
Date Thu, 24 Apr 2008 08:51:21 GMT

    [ https://issues.apache.org/jira/browse/HADOOP-3280?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=12591952#action_12591952

Arun C Murthy commented on HADOOP-3280:

bq. Should "mapred.child.ulimit" be applied to java tasks as well? 

I'm not sure we want to introduce that right-away into 0.17.0.

Medium/long term we could let mapred.child.ulimit be used for passing limits like no. of files,
max. user processes etc. (i.e. -m, -n etc.). Thoughts?

Btw, I can't seem to get TestUlimit to work with 'ulimit -m', Devaraj suggested I rework the
testcase to use a simple streaming script which will just do 'ulimit -m' and check that the
user-defined value is honoured... I'll rework the patch.

bq. You can remove the isCygwin check in streaming also(done already in pipes), as you are
taking care of it in getUlimitMemoryCommand().

No, unfortuanately that check is necessary since streaming doens't use bash in Windows...

> virtual address space limits break streaming apps
> -------------------------------------------------
>                 Key: HADOOP-3280
>                 URL: https://issues.apache.org/jira/browse/HADOOP-3280
>             Project: Hadoop Core
>          Issue Type: Bug
>          Components: contrib/streaming
>            Reporter: Rick Cox
>            Assignee: Arun C Murthy
>            Priority: Blocker
>             Fix For: 0.17.0
>         Attachments: HADOOP-3280_0_20080418.patch, HADOOP-3280_1_20080423.patch, patch-3280.txt,
> HADOOP-2765 added a mandatory, hard virtual address space limit to streaming apps based
on the Java process's -Xmx setting.
> This makes it impossible to run a 64-bit streaming app that needs large address spaces
under a 32-bit JVM, even if one is otherwise willing to dramatically increase the -Xmx setting
without cause. Also, unlike Java's -Xmx limit, the virtual address space limit for an arbitrary
UNIX process does not necessarily correspond to RAM usage, so it's likely to be a relatively
difficult to configure limit.
> 2765 was originally opened to allow an optional wrapper script around streaming tasks,
one use case for which was setting a ulimit. That approach seems much less intrusive and more
flexible than the final implementation. The ulimit can also be trivially set by the streaming
task itself without any support from Hadoop.
> Marking this as an 0.17 blocker because it will break deployed apps and there is no workaround

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

View raw message