[
https://issues.apache.org/jira/browse/CASSANDRA-20129?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=17904308#comment-17904308
]
Caleb Rackliffe edited comment on CASSANDRA-20129 at 12/10/24 5:03 PM:
-----------------------------------------------------------------------
Simple patch up here: [https://github.com/apache/cassandra/pull/3735]
EDIT: CI results here: [^ci_summary.html]
Looks like we hit CASSANDRA-17339, and error withÂ
{{test_speculative_data_request}} that we've seen in CASSANDRA-17232, and a
problem in {{NetstatsBootstrapWithoutEntireSSTablesCompressionStreamingTest}}
that I cannot reproduce locally and should have no relation to this patch.
was (Author: maedhroz):
Simple patch up here: [https://github.com/apache/cassandra/pull/3735]
EDIT: CI results here: [^ci_summary.html]
> Restore optimization in MultiCBuilder around building one clustering
> --------------------------------------------------------------------
>
> Key: CASSANDRA-20129
> URL: https://issues.apache.org/jira/browse/CASSANDRA-20129
> Project: Apache Cassandra
> Issue Type: Bug
> Components: Legacy/Local Write-Read Paths
> Reporter: Caleb Rackliffe
> Assignee: Caleb Rackliffe
> Priority: Normal
> Fix For: 5.x
>
> Attachments: 41 BTree builder inits.png, 41 vs trunk GC.png, 41 vs
> trunk TLAB allocations.png, ci_summary.html, post-patch-allocations-2.jfr,
> pre-patch-allocations.jfr, trunk BTree builder inits.png
>
> Time Spent: 0.5h
> Remaining Estimate: 0h
>
> Back in CASSANDRA-10409, we optimized the single clustering case for
> {{MultiCBuilder}} to produce less garbage. In CASSANDRA-19341, we de-facto
> removed that optimization, and I'm seeing increased allocations in trunk,
> specifically at least double the number of {{BTree}} builders. In the
> write-heavy workload I'm profiling, this jumps up to become the most common
> TLAB allocation. I'm attaching JFR screenshots from 4.1 and trunk that show
> the codepaths that touch this on both versions. We should consider
> reintroducing something like the old {{OneClusteringBuilder}}.
> I used {{WriteBench}} to explore things here, mostly w/ the TRUNCATE
> {{EndOp}}, and made sure commitlog writes, view updates, and
> {{CassandraTableWriteHandler}} tracing were disabled to reduce the noise.
>
> [~blerer]
--
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
