Jeezes - I'm sitting here with red cheeks again.

Replication works fine, I just (completely) misread the docs about the server ID in my.cnf: I must have had something about security in mind or about replication from more than one master, because I read "must be different for each master" where it actually says "must be different for each server".

Now I do have one more question: in earlier releases, there was a remark in the docs saying that you can round-robin replication between two servers and do updates on both, later that remark was removed.

Does anyone know why? Is there, as I suspect, a risk involved with unique keys and/or auto-increment fields? If that's all, I think I could still use it, because updates would always be done on just one server - it's just that which server that is might change from time to time, without any overlap ever. Or am I begging for race conditions around the switchover now?

Regards, Luc.