[ 
https://issues.apache.org/jira/browse/CASSANDRA-12816?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=15605631#comment-15605631
 ] 

Jeremy Hanna commented on CASSANDRA-12816:
------------------------------------------

When [~jjordan] says "it will work", it won't fail.  It will work like it did 
before.  That is, in the scenario where you set auto bootstrap to false when 
you add the new datacenter nodes - since SimpleStrategy ignores datacenter 
information, some of the new nodes will take ownership for some of the ranges 
for the SimpleStrategy keyspace data.  So a rebuild from the source datacenter 
won't get the data that the new datacenter nodes own.  So you won't have all of 
the data from the SimpleStrategy keyspaces when everything is done.  You can 
make sure you don't replicate to the new datacenter in your NTS keyspace config 
and do auto bootstrap = true and actually get the data when the new nodes take 
ownership of those ranges.  However, then you're still left with having to 
migrate SimpleStrategy keyspaces to NetworkTopologyStrategy with two 
datacenters anyway because you really don't want SimpleStrategy with more than 
one datacenter.

This is why you always, *always* change to NetworkTopologyStrategy on all 
non-LocalStrategy keyspaces before adding a second datacenter.

> Rebuild failing while adding new datacenter
> -------------------------------------------
>
>                 Key: CASSANDRA-12816
>                 URL: https://issues.apache.org/jira/browse/CASSANDRA-12816
>             Project: Cassandra
>          Issue Type: Bug
>            Reporter: Jai Bheemsen Rao Dhanwada
>            Priority: Critical
>
> Hello All,
> I have single datacenter with 3 C* nodes and we are trying to expand the 
> cluster to another region/DC. I am seeing the below error while doing a 
> "nodetool rebuild -- name_of_existing_data_center" .  
> {code:java}
> [user@machine ~]$ nodetool rebuild DC1
> nodetool: Unable to find sufficient sources for streaming range 
> (-402178150752044282,-396707578307430827] in keyspace system_distributed
> See 'nodetool help' or 'nodetool help <command>'.
> [user@machine ~]$
> {code}
> {code:java}
> user@cqlsh> SELECT * from system_schema.keyspaces where 
> keyspace_name='system_distributed';
>  keyspace_name | durable_writes | replication
> ---------------+----------------+-------------------------------------------------------------------------------------
>  system_distributed |           True | {'class': 
> 'org.apache.cassandra.locator.SimpleStrategy', 'replication_factor': '3'}
> (1 rows)
> {code}
> To overcome this I have updated system_distributed keyspace to DC1:3 and 
> DC2:3 with NetworkTopologyStrategy
> C* Version - 3.0.8
> Is this a bug that is introduced in 3.0.8 version of cassandra? as I haven't 
> seen this issue with the older versions?



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

Reply via email to