Re: innodb replication

[Marcus Bointon]

On 23 Apr 2008, at 19:04, Rick James wrote:

> Consistent state, yes.  But what about mid-transaction?  Might it be  
> that the snapshot will have half the transaction and have to  
> ROLLBACK, while the live machine (after UNLOCK) will COMMIT the  
> transaction?


A transaction that didn't commit effectively never happened at all as  
far as a slave is concerned, so I'd just expect it to catch up from  
its master, where the transaction did commit. Right?

Marcus
-- 
Marcus Bointon
Synchromedia Limited: Creators of http://www.smartmessages.net/
UK resellers of info@hand CRM solutions
marcus@stripped | http://www.synchromedia.co.uk/