[ https://issues.apache.org/jira/browse/CASSANDRA-9364?page=com.atlassian.jira.plugin.system.issuetabpanels:all-tabpanel ]
Brandon Williams updated CASSANDRA-9364: ---------------------------------------- Attachment: 9364.txt Patch to warn when over cores * 2 + 1024, which covers all the cases of this I've seen. > Add a warning when a high setting for rpc_max_threads is specified for HSHA > --------------------------------------------------------------------------- > > Key: CASSANDRA-9364 > URL: https://issues.apache.org/jira/browse/CASSANDRA-9364 > Project: Cassandra > Issue Type: Improvement > Reporter: Brandon Williams > Fix For: 2.0.15, 2.1.x > > Attachments: 9364.txt > > > In CASSANDRA-7594 we allowed configuring this. The problem is this was > ignored for HSHA previously, so some people go through this scenario: > 1. start with sync and 4000 threads > 2. switch to HSHA, pre-7594 > 3. upgrade to a version after 7594, without changing this setting > 4. experience awful performance due to insane thread contention -- This message was sent by Atlassian JIRA (v6.3.4#6332)