You are viewing a plain text version of this content. The canonical link for it is here.
Posted to common-dev@hadoop.apache.org by "Owen O'Malley (JIRA)" <ji...@apache.org> on 2008/09/19 01:16:44 UTC
[jira] Created: (HADOOP-4208) Shouldn't delete the system dir after
a JobTracker recovery
Shouldn't delete the system dir after a JobTracker recovery
-----------------------------------------------------------
Key: HADOOP-4208
URL: https://issues.apache.org/jira/browse/HADOOP-4208
Project: Hadoop Core
Issue Type: Bug
Reporter: Owen O'Malley
Debugging JobTracker crashes will be easier if the files are preserved rather than deleted on recovery, probably in a sub-directory.
--
This message is automatically generated by JIRA.
-
You can reply to this email to add a comment to the issue online.
[jira] Issue Comment Edited: (HADOOP-4208) Shouldn't delete the
system dir after a JobTracker recovery
Posted by "Owen O'Malley (JIRA)" <ji...@apache.org>.
[ https://issues.apache.org/jira/browse/HADOOP-4208?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=12632443#action_12632443 ]
owen.omalley edited comment on HADOOP-4208 at 9/18/08 4:19 PM:
----------------------------------------------------------------
Also note that checkAndAddJob is also deleting files that don't have the expected pattern of job_ * _ * , which should be removed too.
was (Author: owen.omalley):
Also note that checkAndAddJob is also deleting files that don't have the expected pattern of job_*_*, which should be removed too.
> Shouldn't delete the system dir after a JobTracker recovery
> -----------------------------------------------------------
>
> Key: HADOOP-4208
> URL: https://issues.apache.org/jira/browse/HADOOP-4208
> Project: Hadoop Core
> Issue Type: Bug
> Reporter: Owen O'Malley
>
> Debugging JobTracker crashes will be easier if the files are preserved rather than deleted on recovery, probably in a sub-directory.
--
This message is automatically generated by JIRA.
-
You can reply to this email to add a comment to the issue online.
[jira] Commented: (HADOOP-4208) Shouldn't delete the system dir
after a JobTracker recovery
Posted by "Owen O'Malley (JIRA)" <ji...@apache.org>.
[ https://issues.apache.org/jira/browse/HADOOP-4208?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=12632443#action_12632443 ]
Owen O'Malley commented on HADOOP-4208:
---------------------------------------
Also note that checkAndAddJob is also deleting files that don't have the expected pattern of job_*_*, which should be removed too.
> Shouldn't delete the system dir after a JobTracker recovery
> -----------------------------------------------------------
>
> Key: HADOOP-4208
> URL: https://issues.apache.org/jira/browse/HADOOP-4208
> Project: Hadoop Core
> Issue Type: Bug
> Reporter: Owen O'Malley
>
> Debugging JobTracker crashes will be easier if the files are preserved rather than deleted on recovery, probably in a sub-directory.
--
This message is automatically generated by JIRA.
-
You can reply to this email to add a comment to the issue online.
[jira] Commented: (HADOOP-4208) Shouldn't delete the system dir
after a JobTracker recovery
Posted by "Amar Kamat (JIRA)" <ji...@apache.org>.
[ https://issues.apache.org/jira/browse/HADOOP-4208?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=12632509#action_12632509 ]
Amar Kamat commented on HADOOP-4208:
------------------------------------
Owen,
Pre HADOOP-3245, the whole of system-dir was deleted upon jobtracker (re)start. So if there is nothing to recover, then the jobtracker should switch back to its earlier behavior. Hence anything that doesnt match the pattern will be deleted as done earlier.
> Shouldn't delete the system dir after a JobTracker recovery
> -----------------------------------------------------------
>
> Key: HADOOP-4208
> URL: https://issues.apache.org/jira/browse/HADOOP-4208
> Project: Hadoop Core
> Issue Type: Bug
> Reporter: Owen O'Malley
>
> Debugging JobTracker crashes will be easier if the files are preserved rather than deleted on recovery, probably in a sub-directory.
--
This message is automatically generated by JIRA.
-
You can reply to this email to add a comment to the issue online.
[jira] Commented: (HADOOP-4208) Shouldn't delete the system dir
after a JobTracker recovery
Posted by "Amar Kamat (JIRA)" <ji...@apache.org>.
[ https://issues.apache.org/jira/browse/HADOOP-4208?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=12641387#action_12641387 ]
Amar Kamat commented on HADOOP-4208:
------------------------------------
Job recovery failure is logged to jobtracker's log along with the reason (exception). Deleting the files doesn't cause any problem as there is nothing one can do from those files (job.xml, job.jar and job.split). There is always the history file to see what the final state was. Hence I think this is not a bug but an improvement. Should we move this to 0.20? One question to answer is what all needs to be backed up or preserved and for how long?
> Shouldn't delete the system dir after a JobTracker recovery
> -----------------------------------------------------------
>
> Key: HADOOP-4208
> URL: https://issues.apache.org/jira/browse/HADOOP-4208
> Project: Hadoop Core
> Issue Type: Bug
> Reporter: Owen O'Malley
> Assignee: Amar Kamat
>
> Debugging JobTracker crashes will be easier if the files are preserved rather than deleted on recovery, probably in a sub-directory.
--
This message is automatically generated by JIRA.
-
You can reply to this email to add a comment to the issue online.
[jira] Assigned: (HADOOP-4208) Shouldn't delete the system dir
after a JobTracker recovery
Posted by "Amar Kamat (JIRA)" <ji...@apache.org>.
[ https://issues.apache.org/jira/browse/HADOOP-4208?page=com.atlassian.jira.plugin.system.issuetabpanels:all-tabpanel ]
Amar Kamat reassigned HADOOP-4208:
----------------------------------
Assignee: Amar Kamat
> Shouldn't delete the system dir after a JobTracker recovery
> -----------------------------------------------------------
>
> Key: HADOOP-4208
> URL: https://issues.apache.org/jira/browse/HADOOP-4208
> Project: Hadoop Core
> Issue Type: Bug
> Reporter: Owen O'Malley
> Assignee: Amar Kamat
>
> Debugging JobTracker crashes will be easier if the files are preserved rather than deleted on recovery, probably in a sub-directory.
--
This message is automatically generated by JIRA.
-
You can reply to this email to add a comment to the issue online.