Single-node to multi-node migration

[James Smith]

Hello all,

I have a single-node Scalr system managing about 30 farms of ~250 instances, all mission critical (apparently "this is a POC system, don't put anything mission critical in it" means "put all your mission critical things here, please" to my devs) so I can't just terminate everything. Is there an approved or documented process for moving from a single-node to a multi-node system? The "Production Deployment Recommendations" documentation has a specific order for standing up a new cluster - is that a reasonable order to follow for migrating off of an existing system?

[James Smith]

Trying to start by separating the DB. Decided I was going to cluster the existing DB to a new slave, then make the slave the new master - but when trying to run the kickstart-replication script, I'm getting "Binary logging was not enabled when you boostrapped this MySQL server." I worked around this by manually running 

/opt/scalr-server/embedded/bin/mysqlreplicate --master=root:<root pw>@<master ip>:3306 --slave=root:<root pw>@<slave ip>:3306 --rpl=repl:<repl pw> --pedantic -vv --start-from-beginning