Hi, we have a Percona XtraDB Cluster with 3 nodes on our environment.
We need to create other two Disaster Recovery Percona XtraDB Clusters starting from first Cluster, so we decide to use the MySQL replication from one node of first Cluster to the first node of the second Cluster, we know that we can have that node out from Cluster, but we don’t know other way to do it.
So this is that we would like to have (or if there is another way, we would like to know)
Node 1 (Master) , Node 2 (Master), Node 3 (Master, Master for second Percona Cluster)
Node 1 (Master, Slave of first Percona Cluster) Node 2 (Master) Node 3 (Master, Master for third Percona Cluster)
Node 1 (Master, Slave of second Percona Cluster) Node 2 (Master) Node 3 (Master)
The node 3 of first and second Cluster have
log_slave_updates = 1
else we had inconsistence data replication on the first node of DR Clusters (error on updates records, not found).
After setting up the first node of the second Cluster like Slave, we started Percona XtraDB cluster on the other two nodes and all seemed right, all the 3 nodes were synced, updates seemed to work with no problems.
But we discovered that when we are adding a new user/grant on the first Cluster, it is only replicated on the Slave of the second Cluster, but not on the other Cluster nodes of second Cluster, we have verified that also on the third Cluster no user/grant was created (because data was not existing on the node 3 of second Cluster).
Is there a way to fix it ?
Is it a bug ?
Is there a best practice to do Disaster Recovery on N (2 or mote) Percona XtraDB Clusters ?
Thanks for your time and best regards.