accumulo-notifications mailing list archives

Site index · List index
Message view « Date » · « Thread »
Top « Date » · « Thread »
From "Billie Rinaldi (JIRA)" <j...@apache.org>
Subject [jira] [Commented] (ACCUMULO-507) Large amount of ranges can prevent job from kicking off
Date Fri, 19 Apr 2013 18:33:15 GMT

    [ https://issues.apache.org/jira/browse/ACCUMULO-507?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=13636704#comment-13636704
] 

Billie Rinaldi commented on ACCUMULO-507:
-----------------------------------------

Happily, I have verified that my initial impression was incorrect: the size limit is on the
job configuration file only.

bq. What about random files being left around?

It looks like the files are created in the user's home directory.  I'm not sure how FileSystem.deleteOnExit
works, but since the ACCUMULO-826 bug was about the files being deleted too soon, I'm inclined
to think it's effective.  This is what the javadocs say: "Mark a path to be deleted when FileSystem
is closed. When the JVM shuts down, all FileSystem objects will be closed automatically. Then,
the marked path will be deleted as a result of closing the FileSystem."

I'll look at recreating a patch.  For 1.5.0?
                
> Large amount of ranges can prevent job from kicking off
> -------------------------------------------------------
>
>                 Key: ACCUMULO-507
>                 URL: https://issues.apache.org/jira/browse/ACCUMULO-507
>             Project: Accumulo
>          Issue Type: Improvement
>          Components: client
>    Affects Versions: 1.3.5
>            Reporter: John Vines
>            Assignee: Billie Rinaldi
>            Priority: Minor
>              Labels: mapreduce
>
> We use the various ranges a user provides to create splits. Those get read when the job
is submitted by the client. On the client side, those ranges are used to get all of the splits,
and then the job is started. If the configuration is too large, the job will fail to submit
(this size is configurable, but that's besides the point). We should look into clearing the
ranges out of the jobconf if it's large to prevent this error, since at this point the ranges
are no longer needed in the configuration.

--
This message is automatically generated by JIRA.
If you think it was sent incorrectly, please contact your JIRA administrators
For more information on JIRA, see: http://www.atlassian.com/software/jira

Mime
View raw message