[
https://issues.apache.org/jira/browse/MAPREDUCE-2668?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=13119318#comment-13119318
]
Thomas Graves commented on MAPREDUCE-2668:
------------------------------------------
some background:
Services are started per container. Theoretically a container could have 0 - X
services. The service can't necessarily STOP when the container is finished -
for instance map tasks use the shuffle handler service to serve up the map
output to reduce tasks after mapper is complete.
Right now APPLICATION_STOP isn't sent at all. Since it can't be sent when
container finishes we need to send it when the application finishes. The
application then either needs to know what all services the containers used on
that particular nm or we broadcast the stop message to the services and they
stop as necessary. I've chosen to do the second route.
> MR-279: APPLICATION_STOP is never sent to AuxServices
> -----------------------------------------------------
>
> Key: MAPREDUCE-2668
> URL: https://issues.apache.org/jira/browse/MAPREDUCE-2668
> Project: Hadoop Map/Reduce
> Issue Type: Bug
> Components: mrv2
> Affects Versions: 0.23.0
> Reporter: Robert Joseph Evans
> Assignee: Thomas Graves
> Priority: Blocker
> Fix For: 0.23.0
>
>
> APPLICATION_STOP is never sent to the AuxServices only APPLICATION_INIT.
> This means that all map intermediate data will never be deleted.
--
This message is automatically generated by JIRA.
If you think it was sent incorrectly, please contact your JIRA administrators:
https://issues.apache.org/jira/secure/ContactAdministrators!default.jspa
For more information on JIRA, see: http://www.atlassian.com/software/jira
