Eric, that is very helpful. Thanks. Assuming the master and slave are in sync, is there a reason the checksums would not match? I would rather not dump the database and run an external checksum unless I have to.
-- Eric Robinson Director of Information Technology Physician Select Management, LLC 775.720.2082 -----Original Message----- From: Eric Braswell [mailto:[EMAIL PROTECTED] Sent: Thursday, April 20, 2006 1:49 AM To: Robinson, Eric Cc: mysql@lists.mysql.com Subject: Re: How to Verify Replication Status? There are only a very limited set of circumstances where slaves could get out of sync, and if everything is set up right, it basically should not happen. See: http://dev.mysql.com/doc/refman/5.0/en/replication-rules.html And: http://dev.mysql.com/doc/refman/5.0/en/replication-features.html CHECKSUM TABLE is a good option if it's a read-only table or you can stop writes (or just replication) long enough to do that. It only works on MyISAM. You could also do basically the same thing by dumping the data in the same way on each server, and running a checksum (e.g. md5sum) or diff tool. One thing I have done is to use: mysqldump --skip-opt {database} > {database}.sql ..on each machine, then diffed the files using "diff" (note *nix bias here). Using skip-opt to output inserts on individual lines allows you to compare the data to see exactly where any differences are. But this won't help you if you can't transfer all the data to one place -- you could just do a checksum then on both sides and compare that. Would be pretty easy to script that and perform periodic checks. Eric -- Eric Braswell Web Manager MySQL AB Cupertino, USA Robinson, Eric wrote: > I have master-slave replication working fine. However, I worry about the > possibility of the master and slave accidentally getting out of > synchronization. Are there circumstances (other than a direct INSERT to > the slave) that could cause the master and slave to be out of sync? Is > there a way to periodically do some kind of full check to verify that > the slave is an exact duplicate of the master? I thought of just > counting the rows in all the tables on both servers, but that only tells > part of the story. Is the a more elegant and complete way? Also, the > servers are separated by a slow WAN link, so transferring the whole > database across the network is not an option. > > > > -- > > Eric Robinson > > Director of Information Technology > > Physician Select Management, LLC -- MySQL General Mailing List For list archives: http://lists.mysql.com/mysql To unsubscribe: http://lists.mysql.com/[EMAIL PROTECTED]
