Traditionally when we assign DNS servers to hosts we assign more than one. For example, we may assign 8.8.8.8 as a primary an 8.8.4.4 as a secondary so in the unlikely event that one is ever down, we have a second one to fall back on. When we set up a single Pi-hole server with no failover we lose this redundancy, meaning that if our Pi-hole ever goes down or we want to restart it for updates, we lose DNS until it is back.

The simplest way to regain this redundancy is just to run 2 independent Pi-hole servers and set them as the primary and fallback DNS servers. However, this approach has a couple of disadvantages. The main problem is that we have no setting synchronisation between our servers. If we update the block lists on one, we have to remember to update them on the other. We also can’t guarantee which server clients will connect to, so if one is down, clients may still try to connect and then have to wait for a timeout before querying the second server. This can cause performance drops when one of the servers is offline.

https://davidshomelab.com/pi-hole-failover-with-keepalived/

Pi-hole failover using Gravity Sync and Keepalived - David's Homelab

Use Gravity Sync and IP failover to build a highly available Pi-hole cluster to keep your network online even if a single Pi-hole node fails.

^{David (David's Homelab)}