[jira] [Commented] (HBASE-15146) Don't block on Reader threads queueing to a scheduler queue

["Hiroshi Ikeda (JIRA)" <ji...@apache.org>]

    [ https://issues.apache.org/jira/browse/HBASE-15146?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=15343718#comment-15343718 ] 

Hiroshi Ikeda commented on HBASE-15146:
---------------------------------------

Changing the published method RWQueueRpcExecutor.dispatch etc. breaks compatibility against explicitly declaring the contract by InterfaceAudience.

It doesn't seem to make sense that reader threads hold the initiative from worker threads and continue to just reject incoming requests. Moreover, in my old experience, Selector.select immediately causes a context switch when an event occurs, and this patch might make worse performance in such subtle heavy congestion.

In general, gradually reducing performance is rather preferable in heavy load.

> Don't block on Reader threads queueing to a scheduler queue
> -----------------------------------------------------------
>
>                 Key: HBASE-15146
>                 URL: https://issues.apache.org/jira/browse/HBASE-15146
>             Project: HBase
>          Issue Type: Bug
>    Affects Versions: 1.2.0
>            Reporter: Elliott Clark
>            Assignee: Elliott Clark
>            Priority: Blocker
>             Fix For: 2.0.0, 1.2.0, 1.3.0
>
>         Attachments: HBASE-15146-v7.patch, HBASE-15146-v8.patch, HBASE-15146-v8.patch, HBASE-15146.0.patch, HBASE-15146.1.patch, HBASE-15146.2.patch, HBASE-15146.3.patch, HBASE-15146.4.patch, HBASE-15146.5.patch, HBASE-15146.6.patch
>
>
> Blocking on the epoll thread is awful. The new rpc scheduler can have lots of different queues. Those queues have different capacity limits. Currently the dispatch method can block trying to add the the blocking queue in any of the schedulers.
> This causes readers to block, tcp acks are delayed, and everything slows down.



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