cassandra-commits mailing list archives

Site index · List index
Message view « Date » · « Thread »
Top « Date » · « Thread »
From "Jonathan Ellis (JIRA)" <j...@apache.org>
Subject [jira] [Commented] (CASSANDRA-5455) Remove PBSPredictor
Date Fri, 10 May 2013 19:53:16 GMT

    [ https://issues.apache.org/jira/browse/CASSANDRA-5455?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=13654771#comment-13654771
] 

Jonathan Ellis commented on CASSANDRA-5455:
-------------------------------------------

bq. the PBS latency tracking will record both how long it took for the request to reach a
remote replica and be processed as well as how long the return trip takes.

Hmm, I don't see that happening on the 1.2 branch.  It looks to me like like PBS was trying
to measure raw message RTT (slightly incorrectly since it was clicking Start when we enqueued
the message rather than when we sent).  Which, granted, *is* different from the SP "time to
complete request" metrics.

Would the latter be "close enough?"  I'd rather only track one set of mostly similar metrics,
given the choice.

bq. modifying StorageProxy to log these latencies in addition to the coarser granularity measurements
it already takes

I'm fine with either "provide it as SPMBean methods" or "create a separate MBean that we only
kick off if individual data point collection is enabled."

Should we make collection be a fraction (0..1) rather than on/off?  ISTM that 10% or 1% of
requests could provide enough information on a busy system, and those CLQ objects could become
a contention point w/ enough cores busy.
                
> Remove PBSPredictor
> -------------------
>
>                 Key: CASSANDRA-5455
>                 URL: https://issues.apache.org/jira/browse/CASSANDRA-5455
>             Project: Cassandra
>          Issue Type: Bug
>            Reporter: Jonathan Ellis
>            Assignee: Jonathan Ellis
>             Fix For: 2.0
>
>         Attachments: 5455.txt
>
>
> It was a fun experiment, but it's unmaintained and the bar to understanding what is going
on is high.  Case in point: PBSTest has been failing intermittently for some time now, possibly
even since it was created.  Or possibly not and it was a regression from a refactoring we
did.  Who knows?

--
This message is automatically generated by JIRA.
If you think it was sent incorrectly, please contact your JIRA administrators
For more information on JIRA, see: http://www.atlassian.com/software/jira

Mime
View raw message