[ 
https://issues.apache.org/jira/browse/YARN-1680?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=14157248#comment-14157248
 ] 

Craig Welch commented on YARN-1680:
-----------------------------------

[~airbots] thanks for your updated WIP patch - I've not looked at it 
extensively yet, but at first glance it looks good to me.  On the original 
patch I noticed that there seems to be a facility for blacklisting racks as 
well as nodes, and I was concerned that that needed to be addressed as well.  
It may be in this patch, but it did not look like it to me.  I do think it can 
be without too much difficulty - I think putting the additions (and removals) 
into sets and then checking to see if the node's rack is in the set during the 
node iteration would do the trick (I may be off here, but that looks like it 
would work to me.)

> availableResources sent to applicationMaster in heartbeat should exclude 
> blacklistedNodes free memory.
> ------------------------------------------------------------------------------------------------------
>
>                 Key: YARN-1680
>                 URL: https://issues.apache.org/jira/browse/YARN-1680
>             Project: Hadoop YARN
>          Issue Type: Sub-task
>    Affects Versions: 2.2.0, 2.3.0
>         Environment: SuSE 11 SP2 + Hadoop-2.3 
>            Reporter: Rohith
>            Assignee: Chen He
>         Attachments: YARN-1680-WIP.patch, YARN-1680-v2.patch, 
> YARN-1680-v2.patch, YARN-1680.patch
>
>
> There are 4 NodeManagers with 8GB each.Total cluster capacity is 32GB.Cluster 
> slow start is set to 1.
> Job is running reducer task occupied 29GB of cluster.One NodeManager(NM-4) is 
> become unstable(3 Map got killed), MRAppMaster blacklisted unstable 
> NodeManager(NM-4). All reducer task are running in cluster now.
> MRAppMaster does not preempt the reducers because for Reducer preemption 
> calculation, headRoom is considering blacklisted nodes memory. This makes 
> jobs to hang forever(ResourceManager does not assing any new containers on 
> blacklisted nodes but returns availableResouce considers cluster free 
> memory). 



--
This message was sent by Atlassian JIRA
(v6.3.4#6332)

Reply via email to