[
https://issues.apache.org/jira/browse/CASSANDRA-15037?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=16781382#comment-16781382
]
Nitsan Wakart commented on CASSANDRA-15037:
-------------------------------------------
I have hit this issue in the past, but TBH have never considered fixing it as I
moved away from using the summary result to post processing the HDR histogram
files. The solution in my case has been to use the fixed rate mode with an HDR
log file output and select the period of interest in post processing by
HdrLogProcessing ([https://github.com/nitsanw/HdrLogProcessing).] This method
also allows for unifying multiple load generator outputs and trimming away the
ramp up and ramp down periods where load is uneven or the load generators are
warming up or the server is warming up etc.
Having said all that, I think this is a feature worth implementing for those
who rely on the command line output and summary.
> cassandra-stress doesn't provide warmup for duration based executions
> ---------------------------------------------------------------------
>
> Key: CASSANDRA-15037
> URL: https://issues.apache.org/jira/browse/CASSANDRA-15037
> Project: Cassandra
> Issue Type: Bug
> Components: Legacy/Tools
> Reporter: Sergey Kuksenko
> Priority: Minor
> Labels: stress
> Fix For: 4.0.x
>
> Attachments: stress_warmup.patch
>
>
> cassandra-stress utility performs warmup only when number of operations is
> specified.
> In case of duration based execution it doesn't provide warmup. That leads to
> count startup/warmup induced latency that maybe considered as undesired
> statistics.
> For example latencies difference between warmed and cold execution even for
> 15 minutes may reach to: 17% for 99th percentile, 35% for 99.9th and 55% for
> max latency.
>
> Attached patch contains example how it may be fixed. It was chosen to use 1/5
> of duration time as warmup time, but probably the proper way to do it is to
> add a new option.
>
--
This message was sent by Atlassian JIRA
(v7.6.3#76005)
---------------------------------------------------------------------
To unsubscribe, e-mail: [email protected]
For additional commands, e-mail: [email protected]
