[
https://issues.apache.org/jira/browse/CASSANDRA-20310?page=com.atlassian.jira.plugin.system.issuetabpanels:all-tabpanel
]
Dmitry Konstantinov updated CASSANDRA-20310:
--------------------------------------------
Description:
While assessing another performance question I have noticed suspicious minor
page faults reported in async profiler cpu profile for access to native memory
in NativeCell and NativeClustering:
!image-2025-02-09-10-37-20-057.png|width=300!
!image-2025-02-09-10-37-37-196.png|width=300!
It makes sense to check how jmalloc reuses memory in case of memtables
allocation use case and maybe to play with dirty_decay_ms/muzzy_decay_ms values
to reduce page faults rate..
Related links:
* [https://github.com/jemalloc/jemalloc/wiki/getting-started]
* [https://github.com/jemalloc/jemalloc/blob/dev/TUNING.md]
* [https://github.com/jemalloc/jemalloc/issues/2438]
* [https://github.com/jemalloc/jemalloc/issues/2294]
* KUDU-1366
* [https://github.com/jemalloc/jemalloc/issues/2395]
* [https://www.jabperf.com/how-to-deter-or-disarm-tlb-shootdowns/]
was:
While assessing another performance question I have noticed suspicious minor
page faults reported in async profiler cpu profile for access to native memory
in NativeCell and NativeClustering:
!image-2025-02-09-10-37-20-057.png|width=300!
!image-2025-02-09-10-37-37-196.png|width=300!
It makes sense to check how jmalloc reuses memory in case of memtables
allocation use case and maybe to play with dirty_decay_ms/muzzy_decay_ms values
to reduce page faults rate..
Related links:
* [https://github.com/jemalloc/jemalloc/wiki/getting-started]
* [https://github.com/jemalloc/jemalloc/blob/dev/TUNING.md]
* [https://github.com/jemalloc/jemalloc/issues/2438]
* [https://github.com/jemalloc/jemalloc/issues/2294]
* KUDU-1366
* [https://github.com/jemalloc/jemalloc/issues/2395]
> Analyze page faults for jmalloc allocated memory
> ------------------------------------------------
>
> Key: CASSANDRA-20310
> URL: https://issues.apache.org/jira/browse/CASSANDRA-20310
> Project: Apache Cassandra
> Issue Type: Improvement
> Components: Local/Memtable
> Reporter: Dmitry Konstantinov
> Assignee: Dmitry Konstantinov
> Priority: Low
> Fix For: 5.x
>
> Attachments: 5.1_profile_malloc.html, 5.1_profile_page_faults.html,
> image-2025-02-09-10-37-20-057.png, image-2025-02-09-10-37-37-196.png
>
>
> While assessing another performance question I have noticed suspicious minor
> page faults reported in async profiler cpu profile for access to native
> memory in NativeCell and NativeClustering:
> !image-2025-02-09-10-37-20-057.png|width=300!
> !image-2025-02-09-10-37-37-196.png|width=300!
> It makes sense to check how jmalloc reuses memory in case of memtables
> allocation use case and maybe to play with dirty_decay_ms/muzzy_decay_ms
> values to reduce page faults rate..
> Related links:
> * [https://github.com/jemalloc/jemalloc/wiki/getting-started]
> * [https://github.com/jemalloc/jemalloc/blob/dev/TUNING.md]
> * [https://github.com/jemalloc/jemalloc/issues/2438]
> * [https://github.com/jemalloc/jemalloc/issues/2294]
> * KUDU-1366
> * [https://github.com/jemalloc/jemalloc/issues/2395]
> * [https://www.jabperf.com/how-to-deter-or-disarm-tlb-shootdowns/]
--
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
