[
https://issues.apache.org/jira/browse/DRILL-8132?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=17493915#comment-17493915
]
ASF GitHub Bot commented on DRILL-8132:
---------------------------------------
Leon-WTF opened a new pull request #2464:
URL: https://github.com/apache/drill/pull/2464
# [DRILL-8132](https://issues.apache.org/jira/browse/DRILL-8132):
Improvement of RPC
## Description
As all the connection pairs between drillbits will share the same thread
pool of netty, so I propose to at least sepatate data server from control
client/server. Also changed the default threads number to power of two to
benefit from PowerOfTwoEventExecutorChooser of Netty.
## Documentation
NA
## Testing
Start two or more drillbits, see the query which uses two or more drillbits
to scan can perform correctly.
--
This is an automated message from the Apache Git Service.
To respond to the message, please log on to GitHub and use the
URL above to go to the specific comment.
To unsubscribe, e-mail: [email protected]
For queries about this service, please contact Infrastructure at:
[email protected]
> Improvement of RPC
> ------------------
>
> Key: DRILL-8132
> URL: https://issues.apache.org/jira/browse/DRILL-8132
> Project: Apache Drill
> Issue Type: Improvement
> Components: Execution - RPC
> Reporter: wtf
> Assignee: wtf
> Priority: Minor
>
> # Separate EventLoopGroup between dataserver and control client/server, as
> the node which is busy on receiving data from different nodes may occupy all
> EventLoopGroup threads, will block the control rpc message, e.g. fragment
> transfer.
> # Change the default threads of EventLoopGroup to 8(current is 10), as it
> can improve performance if the number is power of two based on
> PowerOfTwoEventExecutorChooser of Netty.
--
This message was sent by Atlassian Jira
(v8.20.1#820001)
