Mailing-List: contact commits-help@cassandra.apache.org; run by ezmlm
Precedence: bulk
Reply-To: dev@cassandra.apache.org
Date: Wed, 27 Mar 2013 20:03:17 +0000 (UTC)
From: "Jonathan Ellis (JIRA)" <jira@apache.org>
To: commits@cassandra.apache.org
Message-ID: <JIRA.12639433.1364412777273.66324.1364414597565@arcas>
In-Reply-To: <JIRA.12639433.1364412777273@arcas>
References: <JIRA.12639433.1364412777273@arcas>
Subject: [jira] [Updated] (CASSANDRA-5395) Compaction doesn't remove index
 entries as designed
MIME-Version: 1.0
Content-Type: text/plain; charset=utf-8
Content-Transfer-Encoding: 7bit


     [ https://issues.apache.org/jira/browse/CASSANDRA-5395?page=com.atlassian.jira.plugin.system.issuetabpanels:all-tabpanel ]

Jonathan Ellis updated CASSANDRA-5395:
--------------------------------------

    Attachment: 5395-2.txt

second patch attached (on top of the first) that also avoids creating duplicate index entries during PrecompactedRow.merge.  (Calling indexer.update(A, B) would remove the entry for A and add one for B, but since we're compacting we know that an entry for B already exists.)

switched to a merge-column-at-a-time approach similar to what LCR uses.

also tweaked LCR's reducer to short-circuit the column lookup if there is no index involved.
                
> Compaction doesn't remove index entries as designed
> ---------------------------------------------------
>
>                 Key: CASSANDRA-5395
>                 URL: https://issues.apache.org/jira/browse/CASSANDRA-5395
>             Project: Cassandra
>          Issue Type: Bug
>          Components: Core
>    Affects Versions: 1.2.0
>            Reporter: Jonathan Ellis
>            Priority: Minor
>             Fix For: 1.2.4
>
>         Attachments: 5395-2.txt, 5395.txt
>
>
> PerColumnIndexUpdater ignores updates where the new value is a tombstone.  It should still remove the index entry on oldColumn.
> (Note that this will not affect user-visible correctness, since KeysSearcher/CompositeSearcher will issue deletes against stale index entries, but having more stale entries than we "should" could affect performance.)

--
This message is automatically generated by JIRA.
If you think it was sent incorrectly, please contact your JIRA administrators
For more information on JIRA, see: http://www.atlassian.com/software/jira