[
https://issues.apache.org/jira/browse/CASSANDRA-15531?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=17623790#comment-17623790
]
Michael Semb Wever commented on CASSANDRA-15531:
------------------------------------------------
I am not sure I agree with this. We see it sometimes, more often with DSE, but
not enough to warrant a default change from {{`disk_access_mode: auto`}}
Switching to mmap_index_only comes with a cost (higher heap usage), and I
suspect it wiser to only do when warranted.
Docs can be improved to help troubleshoot and document when the change is
warranted.
> Improve docs on disk_access_mode, specifically post CASSANDRA-8464
> ------------------------------------------------------------------
>
> Key: CASSANDRA-15531
> URL: https://issues.apache.org/jira/browse/CASSANDRA-15531
> Project: Cassandra
> Issue Type: Improvement
> Components: Documentation/Website
> Reporter: Erick Ramirez (deprecated)
> Assignee: Erick Ramirez
> Priority: Normal
> Labels: documentation
>
> h2. Background
> [~tsteinmaurer] brought up [on
> Slack|https://the-asf.slack.com/archives/CJZLTM05A/p1580298424288700] that
> they've been hit by the `oom-killer` as a result of SSTables being mmaped.
> h2. Action
> Based on feedback from Thomas Steinmaurer, Jean Carlo & Sam Kearon, I am
> going to push the info I've written in this KB article to the official docs –
> [Increased memory use on nodes after upgrading to DSE 5.0 or DSE
> 5.1|https://support.datastax.com/s/article/Increased-memory-use-on-nodes-after-upgrading-to-DSE-50-or-DSE-51].
> I will revise it specifically for affected OSS versions.
--
This message was sent by Atlassian Jira
(v8.20.10#820010)
---------------------------------------------------------------------
To unsubscribe, e-mail: [email protected]
For additional commands, e-mail: [email protected]
