[
https://issues.apache.org/jira/browse/STORM-1569?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=15163761#comment-15163761
]
ASF GitHub Bot commented on STORM-1569:
---------------------------------------
Github user Parth-Brahmbhatt commented on the pull request:
https://github.com/apache/storm/pull/1144#issuecomment-188447027
@knusbaum I think we should. WE have see RejectedExecutionException on > 5
different customer clusters. I can backport this if others also agree.
> Allowing users to specify the nimbus thrift server queue size.
> --------------------------------------------------------------
>
> Key: STORM-1569
> URL: https://issues.apache.org/jira/browse/STORM-1569
> Project: Apache Storm
> Issue Type: Improvement
> Components: storm-core
> Affects Versions: 0.10.0
> Reporter: Parth Brahmbhatt
> Assignee: Parth Brahmbhatt
> Fix For: 2.0.0
>
>
> Currently the nimbus sever in secure mode uses
> https://docs.oracle.com/javase/7/docs/api/java/util/concurrent/ThreadPoolExecutor.html
> Backed by
> https://docs.oracle.com/javase/7/docs/api/java/util/concurrent/SynchronousQueue.html,
> Please see
> https://github.com/apache/thrift/blob/0.9.2/lib/java/src/org/apache/thrift/server/TThreadPoolServer.java#L132.
> This means that if all executor threads are busy serving a request and new
> requests come in we will see RejectedExecutionExceptions in logs once they
> have reached the retry limit. Instead we should allow the requests to be
> queued. This patch allows the requests to be queued by replacing
> SynchronousQueue with
> https://docs.oracle.com/javase/7/docs/api/java/util/concurrent/ArrayBlockingQueue.html
> with default size of 100000 requests which should be large enough for most
> applications. Applications can modify this default by adding the config
> nimbus.queue.size to their storm.yaml and bouncing nimbus.
--
This message was sent by Atlassian JIRA
(v6.3.4#6332)
