Hyper-converged High Available Homelab with Proxmox.md

Hyper-converged High Available Homelab with Proxmox

This is me documenting my journey moving my Homelab from a Qnap NAS and a Single host Proxmox server to a Hyper-converged multi-node Proxmox Cluster.

The reason to document it here is twofold:

1. Information often it scattered 'all over the place', but never 100% applicable to the setup I have.
2. To remember 'what the fuck' did I do some months ago.
3. Writing it for 'a public' forces me to think it all through again and make sure it's correct.

It's written 'first to scratch my own itch' but hopefully it benefits others too, or even better, that others improve upon my implementations. Feel free to comment or share improvements and insights!

Write-Ups

Proxmox

• High Available Proxmox cluster with Ceph and Dynamic Routing - Setup Networking
• High Available Proxmox cluster with Ceph and Dynamic Routing - Setup Proxmox and Ceph
• High Available Proxmox cluster with Ceph and Dynamic Routing - Managing and Troubleshooting
• Create Erasure Coded CephFS Pools on Proxmox
• Mount Volumes into Proxmox VMs with Virtiofs

Docker

• Docker Swarm in Vm's with CephFS
• Docker Swarm in LXC Containers
• Docker Swarm Keepalived

Applications

• High Available Pi-hole failover cluster using Keepalived and Orbital Sync

---

Pictures

A picture of three out of the four nodes.

Topton S500 Mini PCs

HA Cluster

Ceph Status

What was your "final" config: Swarm in LXCs or Swarm in VMs? What would would choose knowing everything you know now?

Final Config was a Swarm in VMs, since I don't use many containers. I moved away from it and now only use Proxmox VE Helper-Scripts to deploy LXC's. Currently, I'm only running a standalone Docker Host inside CasaOS for two small Docker Containers. Knowing what I know now, I picked the LXC's Helper Scripts over VM's because I'm just running Apps for which that's enough.