In a Puppet Enterprise installation, bringing on new compile masters is very easy. Simply add a new node to the PE Master Classification group underneath the PE Infrastructure:

These nodes will receive the same configuration as the Primary Master, including code manager configuration and necessary connections to PuppetDB. There are no hidden tricks to managing additional compile masters in Puppet Enterprise. Classify and add them to a load balancer.

In open source, we need to ensure each Puppet Master is configured to use PuppetDB. Luckily, puppetlabs/puppetdb provides that connection for us:

class profile::puppetserver {
  class { 'puppetdb::master::config':
    puppetdb_server => <hostname of PuppetDB>,
  }
}

We'll still need to make sure this open source installation has the ability to retrieve code. r10k does not federate across servers, unlike Code Manager, so you'll need to determine a strategy for deploying code out to these masters. One easy method of managing this is included in the puppet/r10k module! Not only can the puppet/r10k module configure r10k in the same way across each Puppetserver, but a new Puppet task is available for deploying code in that module. This can be run from the command line, or preferably from a CI/CD server on commit:

$ puppet task run r10k::deploy environment=production -n puppet-master1,puppet-master2,puppet-master3