[ 
https://issues.apache.org/jira/browse/COUCHDB-141?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=12893749#action_12893749
 ] 

Chris Anderson commented on COUCHDB-141:
----------------------------------------

the challenge with this ticket is determining if 2 databases are copies of the 
same file.

maybe we should think about this differently. if instead of trying to 
automatically suss out which seq num the dbs remain equal up to, we could much 
more easily add a parameter to replication like source_start_seq : N, this 
would start replication from the source at that sequence number.

This can be used to accomplish the goals of this ticket (without any magic) and 
is also generally useful for other scenarios.

If I don't get any push back on this suggestion, I'll rename the ticket in a 
few days. (Or you can, if I forget to.)

> Short-Cut replication initialisation for large databases that are guaranteed 
> to be equal
> ----------------------------------------------------------------------------------------
>
>                 Key: COUCHDB-141
>                 URL: https://issues.apache.org/jira/browse/COUCHDB-141
>             Project: CouchDB
>          Issue Type: New Feature
>          Components: Database Core
>    Affects Versions: 0.9
>            Reporter: Jan Lehnardt
>            Priority: Minor
>
> In a situation where a large database is made available on multiple nodes by 
> copying over the .couch file and where multi-master-replication should keep 
> all nodes in sync, the initial replication in each direction takes quite some 
> time, trying to determine if all databases are in the same state. When we 
> copy the .couch files, they are guaranteed to be equal. It would be nice to 
> be able to tell CouchDB that it doesn't actually have to iterate through all 
> dbs and just accept the fact that we tell it that they are equal.

-- 
This message is automatically generated by JIRA.
-
You can reply to this email to add a comment to the issue online.

Reply via email to