cassandra-commits mailing list archives

Site index · List index
Message view « Date » · « Thread »
Top « Date » · « Thread »
From "Sylvain Lebresne (JIRA)" <j...@apache.org>
Subject [jira] [Commented] (CASSANDRA-10046) RangeTombstone validation missing on thrift
Date Sun, 16 Aug 2015 18:50:46 GMT

    [ https://issues.apache.org/jira/browse/CASSANDRA-10046?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=14698816#comment-14698816
] 

Sylvain Lebresne commented on CASSANDRA-10046:
----------------------------------------------

For the record, this is a regression from CASSANDRA-9704.

This is also not a validation issue, as those test should be passing and we do want to handle
range tombstones whose start differ from their end.

As far as I can tell, the problem is in the {{LegacyLayout.LegacyDeletionInfo.from}} method:
it basically ends up considering all RT as a in-row tombstone, which is definitively incorrect.
And in fact, that methods takes a {{DeletionInfo}}, so it can't have any in-row tombstone
by definition. So I think the simplest way to deal with this is to have {{CassandraServer}}
adds to a {{LegacyDeletionInfo}} object directly (instead of a {{DeletionInfo}}) and have
it deal with in-row tombstones directly. Pushed a branch to do so [here|https://github.com/pcmanus/cassandra/commits/10046].

CI results will be eventually avaible for the [utests|http://cassci.datastax.com/view/Dev/view/pcmanus/job/pcmanus-10046-testall/]
and [dtests|http://cassci.datastax.com/view/Dev/view/pcmanus/job/pcmanus-10046-dtest/].

Marking [~thobbs] as reviewer since he is the original author of CASSANDRA-9704.


> RangeTombstone validation missing on thrift
> -------------------------------------------
>
>                 Key: CASSANDRA-10046
>                 URL: https://issues.apache.org/jira/browse/CASSANDRA-10046
>             Project: Cassandra
>          Issue Type: Bug
>            Reporter: T Jake Luciani
>            Assignee: T Jake Luciani
>            Priority: Minor
>             Fix For: 3.0.0 rc1
>
>
> Currently a number of thrift tests fail due to a NPE
> {code}
> test_range_deletion 
> test_batch_mutate_remove_slice_standard
> test_batch_mutate_remove_slice_of_entire_supercolumns
> {code}
> These errors all stem from the fact we aren't properly throwing a validation exception
when a deletion with a slice that is not start == finish.  
> We should add better validation to throw a clear exception sooner.



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

Mime
View raw message