Hi All, We have a new architecture in place master-master-slave replication Master A and Master B are dual master with log-slave-updates and skip-slave-start on. Slave C is a slave of Master A with log-slave-updates and skip-slave-start on All the updates are going to A and B simultaneously and C is reading all updates from A
A <---> B | C Below is the my.cnf # Master A [mysqld] server-id = 10 log-bin = mysql-bin sync-binlog=1 log-slave-updates replicate-same-server-id = 0 auto_increment_increment = 3 auto_increment_offset = 1 master-host = IP of B master-port = 3306 master-user = repuser master-password = reppass report-host = IP of A event_scheduler = 1 skip-slave-start skip-slave-start #Master B [mysqld] port = 3306 server-id = 20 log-bin = mysql-bin sync-binlog=1 log-slave-updates replicate-same-server-id = 0 auto_increment_increment = 3 auto_increment_offset = 2 master-host = IP of A master-port = 3306 master-user = repuser master-password = reppass report-host = IP of B skip-slave-start event_scheduler = 1 # Master C [mysqld] server-id = 30 log-bin = mysql-bin sync-binlog=1 log-slave-updates replicate-same-server-id = 0 auto_increment_increment = 3 auto_increment_offset = 3 master-host = IP of A master-port = 3306 master-user = repuser master-password = reppass report-host = IP of C skip-slave-start event_scheduler = 1 Recovery Plan: Test cases TEST CASE I: When master A is down, what is the recovery process to promote C as master ? And the design will be C <---> B | A TEST CASE II : When Master B is down,what is the recovery process to follow to promote C as master? And the design will be A <---> C | B TEST CASE III : Worst case when A and B are down,what are the process to follow to recover? Any help would be appreciated. Regards, ./Zakir [Non-text portions of this message have been removed]
