Proxmox Chronicles: From BBQ Sauce Mishaps to Tech Triumphs
Alright so lets say you’re sitting there, BBQ sauce on your HDDS, and a Chic-Fil-A sauce comes flying at your head. What does this have to do with tech? Everything. Chic-Fil-A sauce is versatile, and so is Proxmox.
You may be asking yourself, “did this dude really just compare Proxmox to Chic-Fil-A sauce?”
Yes. Welcome to the madness.
In my lab, I have migrated away from bare metal servers towards using Proxmox for all my hosting needs. (If you are looking for a great Proxmox install guide, here is one of the best guides I have seen). There was a lot of considerations to building out my cluster, though. I needed to have lots of RAM and threads available for current projects, plus the future. So what did I end up getting? I went with a 5 node cluster.
Originally, I was only going to go with a 3 node cluster (more on that later). My first node was something I didn’t expect to have run much of anything, but was more setup so I could have quorum. The first node is an OptiPlex 3060 Micro, running 32gb DDR4 and an i5-8500t. Not the best system ever, but it does the job. My second and third nodes are far more interesting: Two Hyve Zeus V1’s, running 128gb RAM each, and both having two Xeon E5-2695 V2’s. This gives me a nice 24 cores, 48 threads per node, which was more than plenty for what I want to do.
Remember how I said this was only going to be a two node cluster? Well, the good ole USPS happened. I was supposed to take delivery 1 week after ordering, and USPS even had it out for delivery, until the package disappeared. I ended up ordering two other servers, a pair of Dell R620s which I was happier with to be completely honest. I am a fan of Adelle, so this was gonna be perfect (bad memes).
The R620s arrived, and then a day later, the Hyve Zeus’ arrived. I decided to keep all four and full send this server build. For the R620s, I went with a modest 128gb RAM each, along with two Xeon E5-2695 V2’s per server, giving me a nice 16 cores, 32 threads on each node. Like I said, I went full send.
Once I got all of the servers in, I started getting Proxmox installed, which I did have some issues with. The OptiPlex was easy. I’ve applied maybe close to a thousand images on OptiPlex’s in my time, so no issues there. The real struggle came when I was setting up the Hyve Zeus’. If you are attempting this yourself, lookup nomodeset and save yourself from hours of frustration. Once I learned that I needed to set nomodeset, I was able to get Proxmox installed. Of course, what is a homelab without a great naming convention, right? Meet Proxmox1, proxmox2 (Yeah, following your own mediocre naming convention is awesome, especially when you forget to put the p as a capital), and Proxmox3.
Then came the Dell servers. Got those badboys up and running, which wasn’t too bad, just SAS’d it out, ya know. These became Proxmox4 and Proxmox5. Anyways, I’d give my naming convention a good ole Cookshow with Trevor score.
But the question still remains, What do I do when I first get my nodes all up? Cluster them. I joined all of them to the Halsafe cluster (Wow, I can be original with names…) and then setup my NAS as additional storage (more on this in a later post).
The first step was DNS. Had to get that sweet, sweet Pi-Hole running. I decided it was worth not using Ubuntu Server as my OS of choice, but instead I would use Lubuntu 22.04 LTS. I like suffering .-.
I ended up creating two DNS servers. One on proxmox2 named PiHole1 and another on Proxmox3 named PiHole2. (This is an amazing guide on how to get PiHole running. I skipped the Unbound DNS step as I wanted to use Cloudflare.) Since I previously ran PiHole, I had a backup config already saved that contained my blocklists. Once loaded, I was up and running.
Next up was Twingate, my remote connection solution (I will be making a post about Twingate and why it is awesome later on). For Twingate, I wanted to run 3 connectors, so I built out 3 separate VMs on 3 of my nodes. This time, I used Ubuntu Server 24.02 LTS which made it a lot easier. Now, I didn’t need to run 3 different connectors, however, having the redundancy is really nice, especially when you are remote and decide you want to reboot your only VM that has a connector.
There was a handful of other VMs I decided to create as well, but like the kids say, yeet. Those other VMs will be getting their own posts, along with the Twingate and PiHole VMs mentioned in this post.
So there you have it—the whirlwind saga of my Proxmox journey, complete with USPS mishaps, server shenanigans, and plenty of tech triumphs. As I continue to fine-tune my setup and dive deeper into the world of homelabbing, I’ll be sharing more tales from the trenches, along with tips, tricks, and maybe even a meme or two.
Stay tuned for future posts where I’ll delve into topics like Pi-Hole, Twingate, and whatever else the homelab gods throw my way. And hey, if you’ve got any burning questions or just want to swap tech tales, hit me up in the comments or shoot me an email—I’d love to hear from you!
Until next time, keep on techin’ and may your servers stay cool and your BBQ sauce stay off your hard drives.
Catch you on the flip side,
Halsafe 🚀✌️