You are viewing a plain text version of this content. The canonical link for it is here.
Posted to commits@cassandra.apache.org by "Jonathan Ellis (JIRA)" <ji...@apache.org> on 2011/01/18 22:37:43 UTC

[jira] Commented: (CASSANDRA-2004) dynamic snitch doesn't collect local latencies

    [ https://issues.apache.org/jira/browse/CASSANDRA-2004?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=12983392#action_12983392 ] 

Jonathan Ellis commented on CASSANDRA-2004:
-------------------------------------------

going through MS would negatively affect performance, so IMO the "right" solution is to add latency info to the 0.6 code path, then (in another ticket) look at re-optimizing 0.7 the same way.  (0.7 lost this during one of the read repair refactors.)

> dynamic snitch doesn't collect local latencies
> ----------------------------------------------
>
>                 Key: CASSANDRA-2004
>                 URL: https://issues.apache.org/jira/browse/CASSANDRA-2004
>             Project: Cassandra
>          Issue Type: Improvement
>    Affects Versions: 0.6.9
>            Reporter: Brandon Williams
>            Assignee: Brandon Williams
>             Fix For: 0.6.10
>
>
> The dynamic snitch doesn't collect local latencies, which means that if a coordinator has a local replica it will end up preferring whatever the subsnitch returns.  In 0.7 we already collect local latencies by going through MessagingService, so we should do the same in 0.6.

-- 
This message is automatically generated by JIRA.
-
You can reply to this email to add a comment to the issue online.