You are viewing a plain text version of this content. The canonical link for it is here.
Posted to commits@cassandra.apache.org by "Sam Tunnicliffe (JIRA)" <ji...@apache.org> on 2013/10/30 19:35:27 UTC

[jira] [Updated] (CASSANDRA-6272) sstableloader data distribution is broken since 1.2.10

     [ https://issues.apache.org/jira/browse/CASSANDRA-6272?page=com.atlassian.jira.plugin.system.issuetabpanels:all-tabpanel ]

Sam Tunnicliffe updated CASSANDRA-6272:
---------------------------------------

    Attachment: 0001-CASSANDRA-6272-Load-or-build-a-real-IndexSummary-in-.patch

> sstableloader data distribution is broken since 1.2.10
> ------------------------------------------------------
>
>                 Key: CASSANDRA-6272
>                 URL: https://issues.apache.org/jira/browse/CASSANDRA-6272
>             Project: Cassandra
>          Issue Type: Bug
>          Components: Tools
>            Reporter: Sam Tunnicliffe
>            Assignee: Sam Tunnicliffe
>            Priority: Minor
>         Attachments: 0001-CASSANDRA-6272-Load-or-build-a-real-IndexSummary-in-.patch
>
>
> Running sstableloader on 1.2.10+ results in radically different distribution compared with earlier versions. It looks as though the 'bare-bones' IndexSummary created in SSTR.loadForBatch is the cause (CASSANDRA-5555); because it contains only a single entry, we end up with the wrong segment of the index file in SSTR.getPosition (its position is always 0),  so only the first segment of the index file is considered when searching for the range's position.
> This doesn't affect  2.0/trunk



--
This message was sent by Atlassian JIRA
(v6.1#6144)