cassandra-commits mailing list archives

Site index · List index
Message view « Date » · « Thread »
Top « Date » · « Thread »
From "Jeremy Hanna (JIRA)" <j...@apache.org>
Subject [jira] [Commented] (CASSANDRA-4971) Token Generator needs to be partitioner aware
Date Mon, 23 Sep 2013 08:21:02 GMT

    [ https://issues.apache.org/jira/browse/CASSANDRA-4971?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=13774371#comment-13774371
] 

Jeremy Hanna commented on CASSANDRA-4971:
-----------------------------------------

There are still valid reasons to not use vnodes.  If people do want/need to use non-vnodes
and murmur 3, it would be helpful to have a partitioner aware token generator.
                
> Token Generator needs to be partitioner aware
> ---------------------------------------------
>
>                 Key: CASSANDRA-4971
>                 URL: https://issues.apache.org/jira/browse/CASSANDRA-4971
>             Project: Cassandra
>          Issue Type: Bug
>          Components: Tools
>    Affects Versions: 1.2.0 beta 2
>            Reporter: Edward Capriolo
>            Assignee: Edward Capriolo
>            Priority: Minor
>
> I do not agree with the decision to change the default partition from RP to murmur. Also
whatever micro-benchmarking demonstrating performance gains is completely outweighed by disk
latency anyway.
> Assuming we are not going to change the default back there is another issue.
> Historically, there are many blogs and reference material shows how to make tokens for
RandomPartitioner. Indeed the relatively new TokenGenerator is 'unaware' of this change, as
it is giving the user tokens for the Random Partitioner. 
>  
> {noformat}
> [edward@tablitha 2]$ vi /home/edward/.ccm/repository/1.2.0-beta2/tools/bin/token-generator

> [edward@tablitha 2]$ python /home/edward/.ccm/repository/1.2.0-beta2/tools/bin/token-generator

> Token Generator Interactive Mode
> --------------------------------
>  How many datacenters will participate in this Cassandra cluster? 1
>  How many nodes are in datacenter #1? 3
> DC #1:
>   Node #1:                                        0
>   Node #2:   56713727820156410577229101238628035242
>   Node #3:  113427455640312821154458202477256070484
> {noformat}
> This will lead to confusion amount new users and imbalanced rings. We should enhance
the token-generator so it will require input from use on which partitioner they are using
so it can do the appropriate math and give users the correct information.

--
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

Mime
View raw message