Anyone know how replication server id is communicated between nodes? Is it a broadcast?
We have clients with multiple sites and our standard two node cluster deployed to each site has server-id 1 and server-id 2.
The replication connection is private IP VLAN for eth1 for each cluster. The client didn't follow our recommendation and has the replication network going through switches, so occasionally we lose the replication network.
I am trying to determine if I will run into a problem with duplicate server-id's if I create a script to change replication to temporarily go over the production interface (eth0) when we lose the replication connection (eth1) The production interfaces for the various sites can see each other.
At one point a few months ago the replication network for two sister sites was mis-configured and MySQL was throwing errors due to duplicate server id's on the replication network. I am wondering if that was because the server-id is broadcast of if maybe slave 1 (server-id 2) saw slave 2 (server-id 2) due to the misconfiguration.
We have clients with multiple sites and our standard two node cluster deployed to each site has server-id 1 and server-id 2.
The replication connection is private IP VLAN for eth1 for each cluster. The client didn't follow our recommendation and has the replication network going through switches, so occasionally we lose the replication network.
I am trying to determine if I will run into a problem with duplicate server-id's if I create a script to change replication to temporarily go over the production interface (eth0) when we lose the replication connection (eth1) The production interfaces for the various sites can see each other.
At one point a few months ago the replication network for two sister sites was mis-configured and MySQL was throwing errors due to duplicate server id's on the replication network. I am wondering if that was because the server-id is broadcast of if maybe slave 1 (server-id 2) saw slave 2 (server-id 2) due to the misconfiguration.