[
https://issues.apache.org/jira/browse/CASSANDRA-2522?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=13413094#comment-13413094
]
Vladimir Sverzhinsky commented on CASSANDRA-2522:
-------------------------------------------------
It is unclear what pass should we take to track per-cf replicate-on-write
stats.
So far I see two options:
- track reads in CounterMutation.makeReplicationMutation(), this is useful in
a way that we separate r-o-w stats from general read stats in
CFS.getColumnFamily(..);
- we can try to track local writes associated with r-o-w stage
(StorageProxy.counterWriteTask:567) and pass them along to
cm.makeReplicationMutation() which would give us possibility to track
write+read as a whole, but this will require overhead on returning Map<string,
long> from cm.apply();
The choice depends on the way we would like to use r-o-w statistics. What was
your original idea, Stu?
> Record latencies for the replicate-on-write stage
> -------------------------------------------------
>
> Key: CASSANDRA-2522
> URL: https://issues.apache.org/jira/browse/CASSANDRA-2522
> Project: Cassandra
> Issue Type: New Feature
> Components: Core
> Reporter: Stu Hood
> Assignee: Vladimir Sverzhinsky
> Priority: Minor
> Labels: lhf
> Fix For: 1.1.3
>
>
> Like the read and write stages, we should record per-cf latencies for the
> replicate-on-write stage: it doesn't fit cleanly into any of our existing
> monitors.
--
This message is automatically generated by JIRA.
If you think it was sent incorrectly, please contact your JIRA administrators:
https://issues.apache.org/jira/secure/ContactAdministrators!default.jspa
For more information on JIRA, see: http://www.atlassian.com/software/jira
