« Archives in July, 2012

Nice replication setup

If you have a chain replication: A=>B=>C. If B is out of sync for some reason, C will be out of sync and you need to rebuild. The nice thing is that when B is rebuilt, it automatically rebuilds all its slaves (C1, C2, C3, etc.) because B&C are still in replication, the DROP IF EXISTS and those INSERTs will run as normal, which is very nice because you can have a few secondary head slaves (B1, B2, B3) and then hundreds/thousands of real slave (C1,C2,C3,C4,C5). They will cascade data down the chain when you only have to rebuild the secondary masters.