Howdy, I'm trying to add a second slave, slave2, running MySQL 5.0.22 on CentOS 5 to our system that currently has one master and one slave, slave1, running 4.0.24, and somehow slave2 somehow ends up with too many records in many of the 30 tables in the database.
Steps taken: 1. Stopped new records from being inserted into the master, and confirmed with count(*)'s that both master and slave1 were in a static state. 2. Stopped mysqld and commented out in my.cnf the master connection parameters (user, host, password, port) on slave2. 3. Deleted master.info, all mysql-bin and relay-bin files from the mysql data directory on slave2. 4. Deleted all .MYD, .MYI, and .frm files from the replication database directory on slave2. 5. rsync'd the .MYD, .MYI, .frm files from slave1 to slave2. 6. Restarted mysqld on slave2 (now not running as a slave). 7. Confirmed that record counts were consistent across master, slave1 and slave2. 8. Stopped mysqld on slave2, uncommented master connection parameters in my.cnf, and restarted mysqld. 9. Got log file and log position parameters with 'show master status' on the master. 10. Ran 'Change master to... with all fields filled in. 11. Ran 'slave start' on slave2. 12. Rechecked record counts on slave2, and they were too large and out of sync with slave1 and master. I poked around in the data on slave2 and found a number of records had been duplicated, and that accounted for the higher record counts. After starting the application that inserts data into the master, I determined that new records are being inserted correctly into slave2. Seriously out of ideas here. Thanks, David
