Mailing-List: contact commits-help@cassandra.apache.org; run by ezmlm
Precedence: bulk
Reply-To: dev@cassandra.apache.org
Date: Wed, 15 Jun 2016 09:51:09 +0000 (UTC)
From: "Robert Stupp (JIRA)" <jira@apache.org>
To: commits@cassandra.apache.org
Message-ID: <JIRA.12974381.1464736084000.3177.1465984269355@Atlassian.JIRA>
In-Reply-To: <JIRA.12974381.1464736084000@Atlassian.JIRA>
References: <JIRA.12974381.1464736084000@Atlassian.JIRA> <JIRA.12974381.1464736084863@arcas>
Subject: [jira] [Updated] (CASSANDRA-11933) Cache local ranges when
 calculating repair neighbors
MIME-Version: 1.0
Content-Type: text/plain; charset=utf-8
Content-Transfer-Encoding: 7bit
archived-at: Wed, 15 Jun 2016 09:51:11 -0000


     [ https://issues.apache.org/jira/browse/CASSANDRA-11933?page=com.atlassian.jira.plugin.system.issuetabpanels:all-tabpanel ]

Robert Stupp updated CASSANDRA-11933:
-------------------------------------
       Resolution: Fixed
    Fix Version/s: 3.0.8
                   3.8
                   2.2.7
                   2.1.15
           Status: Resolved  (was: Ready to Commit)

Nice stuff!

Committed as 2f74831f4218142f6e118678a3c74c79c1f7e1ed to 2.1 and merged to trunk (cherry-picked from Paulo's branches).

> Cache local ranges when calculating repair neighbors
> ----------------------------------------------------
>
>                 Key: CASSANDRA-11933
>                 URL: https://issues.apache.org/jira/browse/CASSANDRA-11933
>             Project: Cassandra
>          Issue Type: Improvement
>          Components: Core
>            Reporter: Cyril Scetbon
>            Assignee: Mahdi Mohammadi
>             Fix For: 2.1.15, 2.2.7, 3.8, 3.0.8
>
>
> During  a full repair on a ~ 60 nodes cluster, I've been able to see that this stage can be significant (up to 60 percent of the whole time) :
> https://github.com/apache/cassandra/blob/cassandra-2.1/src/java/org/apache/cassandra/service/StorageService.java#L2983-L2997
> It's merely caused by the fact that https://github.com/apache/cassandra/blob/cassandra-2.1/src/java/org/apache/cassandra/service/ActiveRepairService.java#L189 calls {code}ss.getLocalRanges(keyspaceName){code} everytime and that it takes more than 99% of the time. This call takes 600ms when there is no load on the cluster and more if there is. So for 10k ranges, you can imagine that it takes at least 1.5 hours just to compute ranges. 
> Underneath it calls [ReplicationStrategy.getAddressRanges|https://github.com/apache/cassandra/blob/3dcbe90e02440e6ee534f643c7603d50ca08482b/src/java/org/apache/cassandra/locator/AbstractReplicationStrategy.java#L170] which can get pretty inefficient ([~jbellis]'s [words|https://github.com/apache/cassandra/blob/3dcbe90e02440e6ee534f643c7603d50ca08482b/src/java/org/apache/cassandra/locator/AbstractReplicationStrategy.java#L165])
> *ss.getLocalRanges(keyspaceName)* should be cached to avoid having to spend hours on it.


--
This message was sent by Atlassian JIRA
(v6.3.4#6332)