[
https://issues.apache.org/jira/browse/CASSANDRA-16782?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=17393022#comment-17393022
]
Marcus Eriksson commented on CASSANDRA-16782:
---------------------------------------------
Yeah, it might make sense to make this behaviour default for STCS, reason this
patch only adds it to LCS/TWCS is that's the only place we've seen this
bq. might be good enough even without raising the max limit.
The overhead when finishing a compaction is quite large, so running these
1000-sstable compactions really improves performance - especially if we have a
ton of sstables - recalculating the interval tree after each compaction is
quite slow in that case for example.
bq. is there a reason to place the new settings in cassandra.yaml instead of
compaction parameters
We didn't really see a case where we would want to override the defaults for
specific tables on a node - if we make this default for STCS that might need to
change though
> Improve the way we pick sstables for STCS-in-L0 and in TWCS 'current' window
> ----------------------------------------------------------------------------
>
> Key: CASSANDRA-16782
> URL: https://issues.apache.org/jira/browse/CASSANDRA-16782
> Project: Cassandra
> Issue Type: Improvement
> Components: Local/Compaction/LCS, Local/Compaction/TWCS
> Reporter: Marcus Eriksson
> Assignee: Marcus Eriksson
> Priority: Normal
> Fix For: 4.x
>
>
> The goal when being behind in L0 should always be to get the number of
> sstables down to a reasonable level as soon as possible. Currently it is
> common that we run compactions on the large sstables but leave thousands of
> tiny sstables behind.
--
This message was sent by Atlassian Jira
(v8.3.4#803005)
---------------------------------------------------------------------
To unsubscribe, e-mail: [email protected]
For additional commands, e-mail: [email protected]
