You are viewing a plain text version of this content. The canonical link for it is here.

Posted to commits@cassandra.apache.org by "Matt Stump (JIRA)" <ji...@apache.org> on 2014/10/27 18:26:35 UTC

[jira] [Created] (CASSANDRA-8195) Tools for discovering problem/largest partitions

Matt Stump created CASSANDRA-8195:
-------------------------------------

             Summary: Tools for discovering problem/largest partitions
                 Key: CASSANDRA-8195
                 URL: https://issues.apache.org/jira/browse/CASSANDRA-8195
             Project: Cassandra
          Issue Type: Improvement
            Reporter: Matt Stump


A very common problem is that a customer will have one or two very hot partitions. This could be a result of an unforeseen usage pattern or a problem user. Tracking down the offending parition/row currently involves a linear scan of the data. 

We should keep stats on the largest N partitions encountered and make those visible via nodetool as we do with cfhistograms. A reasonable suggestion for N is 10, meaning the top 10 partition keys and their size would be visible. We should also probably keep this info as part of the sstable metadata.



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