A three-node Proxmox cluster at home that I use to run my own services and to learn different Linux distributions.
Each server runs VMs with Debian, Ubuntu and Rocky Linux, which host Docker containers, DNS-level ad blocking and monitoring. Everything is reachable remotely through Tailscale.
The hardest problem so far: one night Uptime Kuma alerted me that a node had frozen completely, and it stayed down until I could hard-reset it. It was the second time it happened. After the first, I had turned off network offloading (TSO/GSO), but that wasn't enough. After ruling out a backup job, a failing disk and ZFS memory usage, I traced it in the kernel logs to the built-in Intel I219-V network card, which has a known hardware bug that makes it hang. That stalled the VMs' network traffic, locked up memory management on the host, and the kernel ended up killing a VM. I disabled Energy Efficient Ethernet, set the host to reboot automatically if it ever locks up again, and installed a dedicated Intel network card.
ProxmoxLinuxDockersystemdTailscale