Thank you for taking the time to comment.

The suggested Pi-hole installation under Proxmox utilises system container technology (LXC), which is more flexible and powerful than using docker.

If you are looking for datacentre grade uptime the best approach is to run Proxmox in a HA (high availability) setup, using three Proxmox nodes. When you update one node the rest of the system carries on as normal. In such a setup your Pi-hole can have 99.9999 uptime assuming the rest of the HW platform is tier-4 datacentre grade.

In homelab scenarios, Proxmox requires reboot when kernel updates take place. If you run production grade Proxmox these updates are infrequent. Even if you update Proxmox frequently, and the DHCP/DNS server goes down say for 5 mins there will be no network disruption unless you have a lot of machines trying to register/renew DHCP leases or DNS queries during this time window.

Proxmox with Pi-hole and any number of VMs/CTs on a decent machine takes less than 1 min to reboot with Pi-hole becoming available almost immediately.

You can configure Pi-hole to be the CT with highest priority on startup / lowest on shutdown, so its unavailability is not noticeable.

The RPi HW platform is not as reliable, especially longer-term, compared to the x86 platform.

Pi-hole under Proxmox, IMHO, is a safer longer term bet than anything you run on the RPi platform. On x86 you have choices and a platform which can scale in both breadth and depth.

RPi is a hobby platform not a production worthy platform, IMHO, particularly if reliability is more important than affordability.

Obviously you can run Docker on Kubernetes or similar container orchestrators, but this is overkill for Pi-hole, again IMHO.