[
https://issues.apache.org/jira/browse/CASSANDRA-19920?page=com.atlassian.jira.plugin.system.issuetabpanels:all-tabpanel
]
Blake Eggleston updated CASSANDRA-19920:
----------------------------------------
Fix Version/s: 5.x
Since Version: 5.x
Source Control Link:
https://github.com/apache/cassandra/commit/3c37d9908dbfd64acc428738e1f9fe0a735294a4
Resolution: Fixed
Status: Resolved (was: Ready to Commit)
> CEP-15 (C*): Read accord repair cfk keys from sstable index.
> ------------------------------------------------------------
>
> Key: CASSANDRA-19920
> URL: https://issues.apache.org/jira/browse/CASSANDRA-19920
> Project: Cassandra
> Issue Type: Bug
> Components: Accord
> Reporter: Blake Eggleston
> Assignee: Blake Eggleston
> Priority: Normal
> Fix For: 5.x
>
> Time Spent: 10m
> Remaining Estimate: 0h
>
> In clusters where accord repairs are dealing with a lot of keys, reading the
> keys via cql can take up to 90% of cpu time. This patch switches to reading
> cfk keys out of the sstable index, which effectively removes that step of
> accord repair from the profiler output. Given the structure of the CFK table
> and how C* reads compound partition keys with local partitioner, this is a
> little involved since we need to properly handle exclusive starts and
> inclusive ends without a full partition key.
--
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
