[
https://issues.apache.org/jira/browse/CASSANDRA-17408?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=17500404#comment-17500404
]
Brandon Williams commented on CASSANDRA-17408:
----------------------------------------------
Attached 10M reads and writes with stress defaults with extra gc logging
(+PrintAdaptiveSizePolicy) and a flamegraph of humongous allocations
(G1CollectedHeap::humongous_obj_allocate). It looks like in 4.0 caffeine does
more HAs for caching for reads, and I find it interesting that for writes 4.0
seems to have more contention on memtable switching which doesn't have an
immediately apparent cause.
> Humongous Allocations after upgrading to 4.0
> --------------------------------------------
>
> Key: CASSANDRA-17408
> URL: https://issues.apache.org/jira/browse/CASSANDRA-17408
> Project: Cassandra
> Issue Type: Bug
> Components: Build
> Reporter: Brandon Williams
> Assignee: Brandon Williams
> Priority: Normal
> Fix For: 4.0.x, 4.x
>
> Attachments: 3.11-gc.log.0.current, 311-10M-read.html,
> 311-10M-write.html, 4.0-gc.log.0.current, 40-10M-read.html, 40-10M-write.html
>
>
> I have seen reports of increased Humongous Allocations after upgrading to 4.0
> when using G1. This is extremely detrimental to how G1 operates and
> seriously affects GC pauses.
--
This message was sent by Atlassian Jira
(v8.20.1#820001)
---------------------------------------------------------------------
To unsubscribe, e-mail: [email protected]
For additional commands, e-mail: [email protected]
