[
https://issues.apache.org/jira/browse/CASSANDRA-7019?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=15166982#comment-15166982
]
Branimir Lambov commented on CASSANDRA-7019:
--------------------------------------------
Implemented a "GarbageCollect" compaction operation that does compaction to
each table separately and takes a {{TombstoneOption}} argument to specify the
granularity of the collection. Exposed as JMX method and nodetool command. The
code is uploaded here:
|[branch|https://github.com/apache/cassandra/compare/trunk...blambov:7019-with-nodtool-command]|[diff
to prev
version|https://github.com/blambov/cassandra/compare/7019-tryouts-no-deserialization...blambov:7019-with-nodetool-command]|[utest|http://cassci.datastax.com/job/blambov-7019-with-nodetool-command-testall/]|[dtest|http://cassci.datastax.com/job/blambov-7019-with-nodetool-command-dtest/]|[cstar_perf|http://cstar.datastax.com/tests/id/0049c06a-db96-11e5-ac91-0256e416528f]
The perf run is trying to see a difference between using the three
{{provide_overlapping_tombstones}} options, as well as the effect of running
{{nodetool garbagecollect}} on later reads. It does appear to show slightly
faster reads during the second stage for the GC-enabled, as well as significant
improvement for reads after both garbage collection passes.
Note: Please disregard {{delete.yaml}}, {{logger.info}} in
{{CompactionController}} and the "Garbage collecting..." printout in
{{GarbageCollect}}. They are temporary, to make sure {{cstar_perf}} is doing
the right thing.
> Improve tombstone compactions
> -----------------------------
>
> Key: CASSANDRA-7019
> URL: https://issues.apache.org/jira/browse/CASSANDRA-7019
> Project: Cassandra
> Issue Type: Improvement
> Components: Compaction
> Reporter: Marcus Eriksson
> Assignee: Branimir Lambov
> Labels: compaction
> Fix For: 3.x
>
>
> When there are no other compactions to do, we trigger a single-sstable
> compaction if there is more than X% droppable tombstones in the sstable.
> In this ticket we should try to include overlapping sstables in those
> compactions to be able to actually drop the tombstones. Might only be doable
> with LCS (with STCS we would probably end up including all sstables)
--
This message was sent by Atlassian JIRA
(v6.3.4#6332)
