[
https://issues.apache.org/jira/browse/CASSANDRA-8633?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=14279285#comment-14279285
]
Aleksey Yeschenko commented on CASSANDRA-8633:
----------------------------------------------
W/ the attached partition4.yaml (1/100/10K) stress just consumes all the cores
w/ generating the batches, and getting nowhere.
> cassandra-stress writes don't respect 'select' parameter
> --------------------------------------------------------
>
> Key: CASSANDRA-8633
> URL: https://issues.apache.org/jira/browse/CASSANDRA-8633
> Project: Cassandra
> Issue Type: Bug
> Reporter: Aleksey Yeschenko
> Assignee: Benedict
> Fix For: 2.1.3
>
> Attachments: partition1.yaml, partition2.yaml, partition3.yaml,
> partition4.yaml
>
>
> W/ the attached profile (partition1.yaml), which has 3 clustering columns,
> each w/ clustering: fixed(100), and select: fixed(1)/1M, stress generates
> huge batches (whereas should only generate single-row mutations).
> {noformat}
> WARN 20:37:59 Batch of prepared statements for [extreme.extreme] is of size
> 269973, exceeding specified threshold of 5120 by 264853.
> {noformat}
> It can be better or worth depending on the distribution of those clustering-s
> (w/ product fixed at 1M).
> W/ 10/1000/100 (partition2.yaml):
> {noformat}
> WARN 20:47:22 Batch of prepared statements for [extreme.extreme] is of size
> 1769445, exceeding specified threshold of 5120 by 1764325.
> {noformat}
> W/ 10k/100/1 (partition3.yaml):
> {noformat}
> WARN 20:50:19 Batch of prepared statements for [extreme.extreme] is of size
> 5373, exceeding specified threshold of 5120 by 253.
> {noformat}
--
This message was sent by Atlassian JIRA
(v6.3.4#6332)
