You are viewing a plain text version of this content. The canonical link for it is here.

Posted to issues@drill.apache.org by "Sudheesh Katkam (JIRA)" <ji...@apache.org> on 2015/10/14 17:48:05 UTC

[jira] [Commented] (DRILL-3242) Enhance RPC layer to offload all request work onto a separate thread.

    [ https://issues.apache.org/jira/browse/DRILL-3242?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=14957116#comment-14957116 ] 

Sudheesh Katkam commented on DRILL-3242:
----------------------------------------

I think this patch is important for higher concurrency. We should not keep a RPC thread busy in any form. Can you rebase on master? (And open a pull request for more comments, since most developers aren't looking at RB anymore?)

> Enhance RPC layer to offload all request work onto a separate thread.
> ---------------------------------------------------------------------
>
>                 Key: DRILL-3242
>                 URL: https://issues.apache.org/jira/browse/DRILL-3242
>             Project: Apache Drill
>          Issue Type: Improvement
>          Components: Execution - RPC
>            Reporter: Chris Westin
>            Assignee: Jacques Nadeau
>             Fix For: 1.3.0
>
>         Attachments: DRILL-3242.patch
>
>
> Right now, the app is responsible for ensuring that very small amounts of work are done on the RPC thread.  In some cases, the app doesn't do this correctly.  Additionally, in high load situations these small amounts of work become no trivial.  As such, we need to make RPC layer protect itself from slow requests/responses.



--
This message was sent by Atlassian JIRA
(v6.3.4#6332)