[ 
https://issues.apache.org/jira/browse/YARN-3655?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=14553636#comment-14553636
 ] 

zhihai xu commented on YARN-3655:
---------------------------------

thanks [~asuresh] for the review. I think the flip-flop won't happen.
bq. At some time T2, the next allocation event (after all nodes have sent 
heartbeat.. or after a continuousScheduling attempt) happens, a reservation of 
2GB is made on each node for appX.
The above reservation won't succeed because maxAMShare limitation.
If it succeeded, then the reservation for appX won't be removed.

thanks [~kasha] for your review. these are great suggestions.
I made the change based on your suggestions. Also I fixed fitsInMaxShare issue 
in this JIRA instead of creating a follow-up JIRA.
I also did some optimizations to remove some duplicate logic.
I find hasContainerForNode already covered getTotalRequiredResources.
If we check hasContainerForNode, then we don't check getTotalRequiredResources.
So I remove getTotalRequiredResources check in assignReservedContainer and 
assignContainer.
Also because okToUnreserve checked hasContainerForNode, we don't need to check 
it again for reserved container in assignContainer.
I uploaded a new patch YARN-3655.002.patch with above change.

> FairScheduler: potential livelock due to maxAMShare limitation and container 
> reservation 
> -----------------------------------------------------------------------------------------
>
>                 Key: YARN-3655
>                 URL: https://issues.apache.org/jira/browse/YARN-3655
>             Project: Hadoop YARN
>          Issue Type: Bug
>          Components: fairscheduler
>    Affects Versions: 2.7.0
>            Reporter: zhihai xu
>            Assignee: zhihai xu
>         Attachments: YARN-3655.000.patch, YARN-3655.001.patch, 
> YARN-3655.002.patch
>
>
> FairScheduler: potential livelock due to maxAMShare limitation and container 
> reservation.
> If a node is reserved by an application, all the other applications don't 
> have any chance to assign a new container on this node, unless the 
> application which reserves the node assigns a new container on this node or 
> releases the reserved container on this node.
> The problem is if an application tries to call assignReservedContainer and 
> fail to get a new container due to maxAMShare limitation, it will block all 
> other applications to use the nodes it reserves. If all other running 
> applications can't release their AM containers due to being blocked by these 
> reserved containers. A livelock situation can happen.
> The following is the code at FSAppAttempt#assignContainer which can cause 
> this potential livelock.
> {code}
>     // Check the AM resource usage for the leaf queue
>     if (!isAmRunning() && !getUnmanagedAM()) {
>       List<ResourceRequest> ask = appSchedulingInfo.getAllResourceRequests();
>       if (ask.isEmpty() || !getQueue().canRunAppAM(
>           ask.get(0).getCapability())) {
>         if (LOG.isDebugEnabled()) {
>           LOG.debug("Skipping allocation because maxAMShare limit would " +
>               "be exceeded");
>         }
>         return Resources.none();
>       }
>     }
> {code}
> To fix this issue, we can unreserve the node if we can't allocate the AM 
> container on the node due to Max AM share limitation and the node is reserved 
> by the application.



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

Reply via email to