[
https://issues.apache.org/jira/browse/CASSANDRA-2020?page=com.atlassian.jira.plugin.system.issuetabpanels:all-tabpanel
]
Brandon Williams updated CASSANDRA-2020:
----------------------------------------
Description: This is due to threads completing towards the end, such that
there aren't enough to fully stress the cluster. The main problem here is that
stress.java is a straight port of stress.py, where each thread runs through
some range until it's done, and the threads finish at different times (probably
offset by jvm warmup time.) Instead, a producer/consumer model would work
better. (was: This is due to threads completing towards the end, such that
there aren't enough to fully stress the cluster. The main problem here is that
stress.java is a straight port of stress.py, where it each thread run through
some range until it's done, and the threads finish at different times (probably
offset by jvm warmup time.) Instead, a producer/consumer model would work
better.)
> stress.java performance falls off heavily towards the end
> ---------------------------------------------------------
>
> Key: CASSANDRA-2020
> URL: https://issues.apache.org/jira/browse/CASSANDRA-2020
> Project: Cassandra
> Issue Type: Improvement
> Reporter: Brandon Williams
> Assignee: Pavel Yaskevich
> Priority: Minor
> Fix For: 0.7.1
>
>
> This is due to threads completing towards the end, such that there aren't
> enough to fully stress the cluster. The main problem here is that
> stress.java is a straight port of stress.py, where each thread runs through
> some range until it's done, and the threads finish at different times
> (probably offset by jvm warmup time.) Instead, a producer/consumer model
> would work better.
--
This message is automatically generated by JIRA.
-
You can reply to this email to add a comment to the issue online.
