[
https://issues.apache.org/jira/browse/CASSANDRA-14796?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=16632504#comment-16632504
]
Chris Lohfink commented on CASSANDRA-14796:
-------------------------------------------
Can be a convenience wrapper over
[forceKeyspaceCompactionForTokenRange|https://github.com/apache/cassandra/blob/trunk/src/java/org/apache/cassandra/service/StorageServiceMBean.java#L286],
you can calculate the token and do this with {{nodetool compact}} with {{-st}}
and {{-et}} for now
> Provide a tool which allows users to force manual compaction by ks.table and
> partition key
> ------------------------------------------------------------------------------------------
>
> Key: CASSANDRA-14796
> URL: https://issues.apache.org/jira/browse/CASSANDRA-14796
> Project: Cassandra
> Issue Type: Wish
> Reporter: Joseph Lynch
> Priority: Minor
>
> It's somewhat common to have a "hot partition" or otherwise "bad partition"
> that is causing neighbors to fail. In addition to CASSANDRA-12106, I think it
> would be really useful to have an optional partition key argument to
> {{nodetool compact}} (similar to {{getendpoints}}) which would determine
> which sstables a partition resides in (similar to {{getsstables}}) and then
> run manual compaction on just those sstables.
> I know at Netflix a common way that we "mitigate" these bad partitions is to
> delete them, but if the partition is really large it would be nice to be able
> to delete them _and_ then force compaction of sstables containing them.
> Obviously {{gc_grace}} get's in the way but it seems like a cheap/useful tool
> either way.
--
This message was sent by Atlassian JIRA
(v7.6.3#76005)
---------------------------------------------------------------------
To unsubscribe, e-mail: [email protected]
For additional commands, e-mail: [email protected]
