[
https://issues.apache.org/jira/browse/TINKERPOP-2406?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=17181324#comment-17181324
]
Divij Vaidya commented on TINKERPOP-2406:
-----------------------------------------
Pull request: https://github.com/apache/tinkerpop/pull/1311
> Delegate processing from event loop to worker threads
> -----------------------------------------------------
>
> Key: TINKERPOP-2406
> URL: https://issues.apache.org/jira/browse/TINKERPOP-2406
> Project: TinkerPop
> Issue Type: Improvement
> Components: driver
> Affects Versions: 3.4.8
> Reporter: Divij Vaidya
> Priority: Minor
>
> Netty client has two thread pools. Event loop thread pool which is
> responsible for handling socket events and another is a worker thread pool
> which is used to delegate busy work from event loop threads so that they can
> continue handling socket events. When programming in Netty, ideally, the
> event loop should never be blocked and must be kept free to respond to events.
> In case of exceptional completion of readComplete which could be triggered by
> a server sending an error code in the response, the event loop thread is used
> to perform the clean up logic which could involve an expensive
> replaceConnection. This blocks the event loop.
--
This message was sent by Atlassian Jira
(v8.3.4#803005)
