[
https://issues.apache.org/jira/browse/CASSANDRA-19715?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=17855947#comment-17855947
]
Venkata Naladala commented on CASSANDRA-19715:
----------------------------------------------
So, with compaction_window_entropy and compaction_window_entropy_unit, we can
ensure that compactions don't happen on all the nodes at the same time?
> TWCS - Provide a flag to pick a time when compaction happens for the n-1
> window
> -------------------------------------------------------------------------------
>
> Key: CASSANDRA-19715
> URL: https://issues.apache.org/jira/browse/CASSANDRA-19715
> Project: Cassandra
> Issue Type: New Feature
> Reporter: Venkata Naladala
> Priority: Normal
>
> Below are the settings for the TWCS. With the settings below, the compaction
> for the n-1 window happens at around 00:00 UTC on all the nodes in the
> cluster. This causes a spike in IO utilization on the entire cluster, which
> in turn increases the read latencies.
> {{{}compaction = { 'class' :
> 'org.apache.cassandra.db.compaction.TimeWindowCompactionStrategy',
> 'compaction_window_size' : 1, 'compaction_window_unit' : 'DAYS',
> 'max_threshold' : 32, 'min_threshold' : 4{}}}}
> Requesting to provide a new option for TWCS where we can specify when the
> compaction for n-1 window happens. This way, we can provide a custom time
> instead of the default time(beginning of the new window) so that we can pick
> low-traffic hours and avoid impacting the customer.
--
This message was sent by Atlassian Jira
(v8.20.10#820010)
---------------------------------------------------------------------
To unsubscribe, e-mail: commits-unsubscr...@cassandra.apache.org
For additional commands, e-mail: commits-h...@cassandra.apache.org
