Hi,
This is more of a question about best practice for setting up the DNS in my home-lab. For a server that hosts DNS (in this case pi-hole with unbound), what should that server's DNS be?
I was thinking of spinning up a third DNS so none of the servers that host DNS reference itself for DNS resolution. I am most concerned that if I turn off my servers for a period of time (like if i go on vacation), it would cause issues with internet connections when they are turned back on. Here is what I'm thinking of doing:
server 1 - primary dns: server2, secondary dns server3
server 2 - primary dns: server1, secondary dns, server3
server 3 - primary dns: server1, secondary dns: server2
I know this may be overkill, or my logic not be correct, but I am using this as a way to learn more about setting up DNS in an environment that is self sufficient and also private.
First, there's really no such thing as primary and secondary DNS. Just about every operating system you will run in to will use all the DNS servers available to them. Some will fire off queries to all of the DNS servers and then favor the one that responds the fastest and then check again later to make sure the speed of replies have not changed. Some will use a round-robin approach.
It is a good idea to not have a DNS server use itself for name resolution. And I haven't seen a situation where a server that was offline for what ever reason ends up having problems when it comes back online.
Since you have unbound running, I'd use that as the servers DNS. unbound is very capable of resolving from a cold state. It may be a little slow as it builds up the cache and trust relationships but it will still function.
The other option would be to have the /etc/resolv.conf of the servers point to 1.1.1.1 or 8.8.8.8 or 9.9.9.9 etc... There really isn't a reason for a server to need filtered DNS queries.
The core value of Pi-hole is to be an educational community. This is the place to learn overkill and then figure out how much of that is really needed. Ask any questions you have and we'll try our best to answer them.