You are viewing a plain text version of this content. The canonical link for it is here.
Posted to mapreduce-issues@hadoop.apache.org by "Arun C Murthy (JIRA)" <ji...@apache.org> on 2009/09/30 21:01:24 UTC
[jira] Created: (MAPREDUCE-1049) Cause for failures due to errors
in job-initialization are not communicated back to the client
Cause for failures due to errors in job-initialization are not communicated back to the client
----------------------------------------------------------------------------------------------
Key: MAPREDUCE-1049
URL: https://issues.apache.org/jira/browse/MAPREDUCE-1049
Project: Hadoop Map/Reduce
Issue Type: Bug
Components: client, jobtracker
Affects Versions: 0.20.1
Reporter: Arun C Murthy
Fix For: 0.22.0
The cause for failures due to errors in job-initialization are not communicated back to the client e.g. job which is terminated for crossing the mapred.jobtracker.maxtasks.per.job limit.
--
This message is automatically generated by JIRA.
-
You can reply to this email to add a comment to the issue online.
[jira] Commented: (MAPREDUCE-1049) Cause for failures due to errors
in job-initialization are not communicated back to the client
Posted by "Vinod K V (JIRA)" <ji...@apache.org>.
[ https://issues.apache.org/jira/browse/MAPREDUCE-1049?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=12762613#action_12762613 ]
Vinod K V commented on MAPREDUCE-1049:
--------------------------------------
It will be difficult to pass it directly back to client because initialization happens asynchronously. May be we can do this using diagnostic messages and let JobClient scan these, similar to task-failures.
This in many ways is similar to job failures due to many other reasons - for e.g, I often find myself scanning through each failed task to see how many times it failed and whether this task's is the reason for the job's failure.
A common solution can be a single job-level diagnostic message that can quote a simple single line reason for failure. This can be used by JobClient on job-failure and by UI too.
> Cause for failures due to errors in job-initialization are not communicated back to the client
> ----------------------------------------------------------------------------------------------
>
> Key: MAPREDUCE-1049
> URL: https://issues.apache.org/jira/browse/MAPREDUCE-1049
> Project: Hadoop Map/Reduce
> Issue Type: Bug
> Components: client, jobtracker
> Affects Versions: 0.20.1
> Reporter: Arun C Murthy
> Fix For: 0.22.0
>
>
> The cause for failures due to errors in job-initialization are not communicated back to the client e.g. job which is terminated for crossing the mapred.jobtracker.maxtasks.per.job limit.
--
This message is automatically generated by JIRA.
-
You can reply to this email to add a comment to the issue online.
[jira] Commented: (MAPREDUCE-1049) Cause for failures due to errors
in job-initialization are not communicated back to the client
Posted by "Amar Kamat (JIRA)" <ji...@apache.org>.
[ https://issues.apache.org/jira/browse/MAPREDUCE-1049?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=12771868#action_12771868 ]
Amar Kamat commented on MAPREDUCE-1049:
---------------------------------------
MAPREDUCE-343 might help solve some part of this.
> Cause for failures due to errors in job-initialization are not communicated back to the client
> ----------------------------------------------------------------------------------------------
>
> Key: MAPREDUCE-1049
> URL: https://issues.apache.org/jira/browse/MAPREDUCE-1049
> Project: Hadoop Map/Reduce
> Issue Type: Bug
> Components: client, jobtracker
> Affects Versions: 0.20.1
> Reporter: Arun C Murthy
> Fix For: 0.22.0
>
>
> The cause for failures due to errors in job-initialization are not communicated back to the client e.g. job which is terminated for crossing the mapred.jobtracker.maxtasks.per.job limit.
--
This message is automatically generated by JIRA.
-
You can reply to this email to add a comment to the issue online.