[infinispan-issues] [JBoss JIRA] (ISPN-3386) Cancel Map/Reduce tasks before they cause OOM errors

[William Burns (JIRA)]

    [ https://issues.jboss.org/browse/ISPN-3386?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=13179253#comment-13179253 ] 

William Burns commented on ISPN-3386:
-------------------------------------

This could also be an issue for distributed streams.

> Cancel Map/Reduce tasks before they cause OOM errors
> ----------------------------------------------------
>
>                 Key: ISPN-3386
>                 URL: https://issues.jboss.org/browse/ISPN-3386
>             Project: Infinispan
>          Issue Type: Enhancement
>          Components: Distributed Execution and Map/Reduce
>    Affects Versions: 5.3.0.Final
>            Reporter: Alan Field
>
> Currently, a MapReduceTask that generates an OutOfMemory error will break a clustered cache. The node where the OOM occurs will generate SuspectExceptions on the other nodes in the cluster.
> In order to avoid this situation, Infinispan should monitor free memory and cancel the MapReduceTask if free memory is below a user definable threshold. [~pruivo] has also suggested allowing the API user to set the priority of tasks, so that the Map Reduce Manager can cancel lower priority tasks first. This functional could also be included with ISPN-2991.



--
This message was sent by Atlassian JIRA
(v6.4.11#64026)