[jira] [Commented] (QPID-7819) [Java Broker] The executor in asynchronous store recoverer is effectively unbounded and might cause running out of threads

classic Classic list List threaded Threaded
1 message Options
Reply | Threaded
Open this post in threaded view
|

[jira] [Commented] (QPID-7819) [Java Broker] The executor in asynchronous store recoverer is effectively unbounded and might cause running out of threads

JIRA jira@apache.org

    [ https://issues.apache.org/jira/browse/QPID-7819?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=16044277#comment-16044277 ]

Rob Godfrey commented on QPID-7819:
-----------------------------------

Is the number of available processors really the right metric?  I'm not sure that store recovery is necessarily CPU intensive - what we actually want to achieve is that all queues are recovering concurrently rather than some queues are blocking others.  I think if you want to limit the number of executors it would make sense to change the processing logic such that the work is chunked up, so recovering of one queue is interleaved with another.

> [Java Broker] The executor in asynchronous store recoverer is effectively unbounded and might cause running out of threads
> --------------------------------------------------------------------------------------------------------------------------
>
>                 Key: QPID-7819
>                 URL: https://issues.apache.org/jira/browse/QPID-7819
>             Project: Qpid
>          Issue Type: Bug
>          Components: Java Broker
>    Affects Versions: 0.32, qpid-java-6.0.7, qpid-java-6.1.3
>            Reporter: Alex Rudyy
>
> The executor service used in asynchronous message store recoverer is bounded by {{Integer.MAX_VALUE}}. On  brokers with thousands of queues and low max process limit (imposed by OS), it might cause Broker to run out of threads. I think we should bound the maximum number of threads by number of available processors.



--
This message was sent by Atlassian JIRA
(v6.3.15#6346)

---------------------------------------------------------------------
To unsubscribe, e-mail: [hidden email]
For additional commands, e-mail: [hidden email]