You are viewing a plain text version of this content. The canonical link for it is here.
Posted to commits@cassandra.apache.org by "T Jake Luciani (JIRA)" <ji...@apache.org> on 2015/05/11 15:26:58 UTC

[jira] [Comment Edited] (CASSANDRA-9282) Warn on unlogged batches

    [ https://issues.apache.org/jira/browse/CASSANDRA-9282?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=14537928#comment-14537928 ] 

T Jake Luciani edited comment on CASSANDRA-9282 at 5/11/15 1:25 PM:
--------------------------------------------------------------------

bq.  do we not handle this correctly in SelectStatement and/or StorageProxy?

The following creates two mutations:
{code}
 BEGIN BATCH  
     update test1.foo set f2 = 'a' where f1 = 'a' 
     update test2.foo set f2 = 'a' where f1 = 'a'  
APPLY BATCH;
{code}

IMutation has both .keyspace() and .key(), so there isn't really a way to reduce this further.

Also, we send these over the wire internally as two separate mutations (even to the same replicas). So it makes to keep the batch log. Since one could be processed and the other not.




was (Author: tjake):
bq.  do we not handle this correctly in SelectStatement and/or StorageProxy?

The following creates two mutations:
{code}
 BEGIN BATCH  
     update test1.foo set f2 = 'a' where f1 = 'a' 
     update test2.foo set f2 = 'a' where f1 = 'a'  
APPLY BATCH;
{code}

IMutation has both .keyspace() and .key(), so there isn't really a way to join them.

Also, we send these over the wire internally as two separate mutations (even to the same replicas). So it makes to keep the batch log. Since one could be processed and the other not.



> Warn on unlogged batches
> ------------------------
>
>                 Key: CASSANDRA-9282
>                 URL: https://issues.apache.org/jira/browse/CASSANDRA-9282
>             Project: Cassandra
>          Issue Type: Bug
>          Components: API
>            Reporter: Jonathan Ellis
>            Assignee: T Jake Luciani
>             Fix For: 2.1.x
>
>
> At least until CASSANDRA-8303 is done and we can block them entirely, we should log a warning when unlogged batches across multiple partition keys are used.  This could either be done by backporting NoSpamLogger and blindly logging every time, or we could add a threshold and warn when more than 10 keys are seen.



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