cassandra-commits mailing list archives

Site index · List index
Message view « Date » · « Thread »
Top « Date » · « Thread »
From "Stefania (JIRA)" <j...@apache.org>
Subject [jira] [Commented] (CASSANDRA-9978) Split/Scrub tools no longer remove original sstable
Date Wed, 05 Aug 2015 08:24:04 GMT

    [ https://issues.apache.org/jira/browse/CASSANDRA-9978?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=14654997#comment-14654997
] 

Stefania commented on CASSANDRA-9978:
-------------------------------------

The problem with the standalone splitter is that we obsolete the originals too late, after
the transaction is committed. We could just swap the split() and obsoleteOriginals() calls
but I prefer to modify compaction task and compaction aware writers to receive a {{keepOriginals}}
boolean in addition to {{isOffline}}, with additional constructors where both are false by
default. This way we can specify when we need to delete originals even if offline rather than
mixing the two together.
 
The problem with scrubbing is an NPE caused by the fact that an offline tracker has no cf
store.

Patch available [here|https://github.com/stef1927/cassandra/tree/9978-3.0].

CI pending:

http://cassci.datastax.com/view/Dev/view/stef1927/job/stef1927-9978-3.0-testall/
http://cassci.datastax.com/view/Dev/view/stef1927/job/stef1927-9978-3.0-dtest/

[~benedict] do you have time to review this one? I plan to observe CI for a bit and submit
a patch tomorrow or shortly afterwards.


> Split/Scrub tools no longer remove original sstable
> ---------------------------------------------------
>
>                 Key: CASSANDRA-9978
>                 URL: https://issues.apache.org/jira/browse/CASSANDRA-9978
>             Project: Cassandra
>          Issue Type: Bug
>            Reporter: T Jake Luciani
>            Assignee: Stefania
>            Priority: Minor
>             Fix For: 3.0 beta 1
>
>
> Looks like CASSANDRA-7066 broke the scrub and split tools.  The orig sstable is no longer
removed.
> I fixed the sstablesplit dtest so you should see the issue now.  The max sstable size
doesn't match expected because it's the original 



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

Mime
View raw message