[ 
https://issues.apache.org/jira/browse/HBASE-16341?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=15419258#comment-15419258
 ] 

Sean Busbey commented on HBASE-16341:
-------------------------------------

generally I would consider changing a default behavior to be incompatible. But 
if this is a significant hit and the variable is configurable, then I'd be fine 
with including it along with an appropriate release note that explains the 
tradeoff and says how to restore the prior behavior.

> Missing bit on "Regression: Random Read/WorkloadC slower in 1.x than 0.98"
> --------------------------------------------------------------------------
>
>                 Key: HBASE-16341
>                 URL: https://issues.apache.org/jira/browse/HBASE-16341
>             Project: HBase
>          Issue Type: Bug
>          Components: rpc
>            Reporter: stack
>            Assignee: stack
>         Attachments: HBASE-16341.patch, HBASE-16341.patch
>
>
> [~larsgeorge] found a missing bit in HBASE-15971 "Regression: Random 
> Read/WorkloadC slower in 1.x than 0.98" Let me fix here. Let me quote the man:
> {code}
> BTW, in constructor we do this
> ```    String callQueueType = conf.get(CALL_QUEUE_TYPE_CONF_KEY,
>         CALL_QUEUE_TYPE_FIFO_CONF_VALUE);
> ```
> (edited)
> [8:19]  
> but in `onConfigurationChange()` we do
> ```    String callQueueType = conf.get(CALL_QUEUE_TYPE_CONF_KEY,
>       CALL_QUEUE_TYPE_DEADLINE_CONF_VALUE);
> ```
> (edited)
> {code}



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

Reply via email to