[ 
https://issues.apache.org/jira/browse/YARN-2314?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=14071726#comment-14071726
 ] 

Jason Lowe commented on YARN-2314:
----------------------------------

I suppose we could use a wait timeout.  I was just matching the behavior when 
it tries to refresh the NM token on an in-use proxy which also waits 
indefinitely.  What's the proposed behavior when the timeout expires?  Log a 
message and then...?  Arguably the timeouts should be on the RPC calls rather 
than the proxy cache, since I'm assuming if we're not willing to wait forever 
for a proxy to be freed up we're also not willing to wait forever for a remote 
call to complete.

> ContainerManagementProtocolProxy can create thousands of threads for a large 
> cluster
> ------------------------------------------------------------------------------------
>
>                 Key: YARN-2314
>                 URL: https://issues.apache.org/jira/browse/YARN-2314
>             Project: Hadoop YARN
>          Issue Type: Bug
>          Components: client
>    Affects Versions: 2.1.0-beta
>            Reporter: Jason Lowe
>            Priority: Critical
>         Attachments: nmproxycachefix.prototype.patch
>
>
> ContainerManagementProtocolProxy has a cache of NM proxies, and the size of 
> this cache is configurable.  However the cache can grow far beyond the 
> configured size when running on a large cluster and blow AM address/container 
> limits.  More details in the first comment.



--
This message was sent by Atlassian JIRA
(v6.2#6252)

Reply via email to