I have setup MYSQLHA as per https://kublr.com/blog/setting-up-mysql-replication-clusters-in-kubernetes-2/ have two nodes up and ready able to deploy pods on each of them and replicate data from master to slave within seconds. 1 Master node 2 Slave nodes VMWARE ESXi setup 3 VM's on seperate subnets
I also have NFS shared setup just in case required.
Ref:- https://kublr.com/blog/setting-up-mysql-replication-clusters-in-kubernetes-2/
How to perform auto fail-overs and scaling?
Async master-slave replication of MySQL is not the best fit for this. I would go for something like Galera replication where all the nodes are active in the cluster, can act as seed nodes for new joiners when you scale up and a simple readiness probe is enough to exclude faulty node / include new ones in the Galera cluster.
Asynchronous replication with master-slave is a good choice for cases that are ie. geographically distributed, so that the latency does not affect your workloads.