[
https://issues.apache.org/jira/browse/CASSANDRA-16333?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=17760978#comment-17760978
]
Branimir Lambov commented on CASSANDRA-16333:
---------------------------------------------
I don't think this describes it correctly.
When compacting data, it will check if a row is deleted by some tombstone in a
newer table that does not participate in the compaction. If it is, it will drop
the row from the result. If this manages to result in the partition completely
being removed from the result of the compaction, this will make the tombstone
that deletes the row purgeable.
To be honest, I had hopes that this can help solve the tombstones problem when
I wrote the code, but now I'm not very confident that it is worth using. When
faced with accumulation of tombstones, performing a major compaction or
switching to LCS or levelled UCS is a better option. Ultimately the problem
should be solved by something that removes the tombstone factor from queries,
and we have something in mind as the long-term solution.
> Document provide_overlapping_tombstones compaction option
> ---------------------------------------------------------
>
> Key: CASSANDRA-16333
> URL: https://issues.apache.org/jira/browse/CASSANDRA-16333
> Project: Cassandra
> Issue Type: Improvement
> Components: Documentation
> Reporter: Paulo Motta
> Assignee: Sumanth Pasupuleti
> Priority: Normal
>
> This option was added on CASSANDRA-7019 but it's not documented. We should
> add it to
> https://cassandra.apache.org/doc/latest/operating/compaction/index.html.
--
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
