I mostly want to make sure I am setting this up correctly.
Expected Behaviour:
Add two conditional forwarders for redundancy. For example I have three piholes. Each with a version of these entries.
true,10.0.0.0/8,10.10.10.8,home.lan
true,10.0.0.0/8,10.10.10.9,home.lan
I have two windows server DNS servers with DHCP in sync for redundancy. I would like for the piholes to use either both, or one or the other depending on the first entry in the list and if it cannot contact one server it will contact the other.
$ dig +noall @localhost test.domain
;; communications error to 127.0.0.1#53: timed out
;; communications error to 127.0.0.1#53: timed out
;; communications error to 127.0.0.1#53: timed out
;; no servers could be reached
$ sudo pihole tail
[..]
18:33:33: query[A] test.domain from 127.0.0.1
18:33:33: forwarded test.domain to 10.10.10.10
18:33:33: forwarded test.domain to 10.10.10.20
I believe queries are distributed according to below logic "Improve detection algorithm ...":
If its not working for you, better upload a debug log and post the resulting token URL here for the mods/devs to inspect!
I may have configured it wrong. I was having some local name resolution problems where the piholes were holding onto some old client names that were no longer correct. I made this thread to make sure if what I wanted it to do would work as I desired.
I'll have to look into it. At least for the two esxi hosted piholes for sure and then use the compute stick hosted one as the secondary for clients. I've seen a few different HA designs people have come up with however most of those involved some middleman instance that would be a single point of failure again. This one seems like it would work like a cluster and would allow instant failover takeover and high redundancy. I'd have to move some IPs around though since I quite like quad 10s as my primary DNS server IP.
I asked not only to educate me, but also the OP.
It was a bit weird/not complete to offer a suggestion but not explaining why.
I've had to perform maintenance on my two Pi-hole nodes a couple of times now, taking one down, and did not experience the issues you describe.
If its not necessary to add complexity, my opinion is dont do it ... KISS.
Plus that link of yours is an active-passive setup:
Meaning only one of your Pi-hole nodes is active processing queries while the other(s) do nothing.
I prefer my current active-active setup which spreads the load plus having redundancy.
While admining a couple of active-passive cluster for work, I've experienced the active node going down, and the passive standby node failing to come on-line for some reason ... causing the whole cluster to fail.
Same for hot-spare disks in a RAID setup.
