--On August 6, 2006 2:01:15 PM -0700 Enrique Sanchez Vela <[EMAIL PROTECTED]> wrote:
One thing that you could do, is to setup a replicating slave server, freeze the replication, perform backups (to tape/disks etc), then re-enable the replication activity, this way you provide a non-stop service without compromising your data availabilty.
I've tried this in the past. Replication on large, non-single-application database servers is hard. When you have over a thousand users on a single (very large) DB server the chances that one of them will trip one of the 'does not replicate correctly to slave and causes slave to stop dead in tracks' bugs is pretty high. We found (even with 4.1 series) that we were restarting or dealing with stopped replication on a daily basis. We stopped trying somewhere around 4.1.11 so things may have improved, but they may not have.
For a single application or a limited set of applications it seems to work great and you can whittle out any cases where the app does something that doesn't replicate correctly somehow.
With large numbers of apps/people using it we usually had to turn down or turn off the error checking on the slave so it wouldn't stop in it's tracks all the time. This of course caused the issue of you were never totally sure that what was on the slave was what was on the master.
-- MySQL General Mailing List For list archives: http://lists.mysql.com/mysql To unsubscribe: http://lists.mysql.com/[EMAIL PROTECTED]
