[Question posted by a user on YugabyteDB Community Slack]
I’m running YugabyteDB 2.11. I’m trying to add another two master nodes to an existing cluster but no matter what I do it gets stuck on PRE_VOTER and never initializes.
Currently, I have 3 masters, 3 tservers, +2 tserver read-replica. Trying to add 2 more masters and tservers. Trying to setup a global YugabyteDB read-only cluster with async-write.
The end goal is to have two masters in NA, two in APAC, two in EU, but it keeps getting stuck in PRE_VOTER when we add a new one.
At the moment, all masters are in the same place. I would like to move masters into diverse locations so that they are as reliable as the cluster itself. changing RF size does not seem to change the PRE_VOTER stuck issue
Any help is appreciated!
Note that you don't need yb-masters on the read replica side. Only on the primary cluster: https://docs.yugabyte.com/latest/deploy/multi-dc/read-replica-clusters/
You can only have as many masters as the set replication factor on the primary cluster https://docs.yugabyte.com/latest/deploy/checklist/#replication
So in your case, you want 1 yb-master in NA, APAC and EU each. Because you’re using 3 replication factor.
The masters should be booted into shell mode (no
--master_address) before they can be added to the cluster. And then it should work.