I have a Synology DS418 with 3x 8TB drives configured for (Synology Hybrid RAID - 1 drive fault tolerance) using Btrfs, providing about 14TB of usable storage that I have about 40% full. I use it as a dumping ground for files, ad-hoc backups, family photo/video archives and such. Some stuff is more organized than others, and I haven’t set it up for off site backups directly from the NAS since I moved everything to it the end of 2021 where everything that was moved to it has an off site backup. I don’t have a lot of data that changes or is added to it, and have so far just used it for a lot of data archival to get the data off some other system and make it accessible better than some floating external drive.

My local network is made up of all UniFi gear with a USG, CloudKey Gen2, some switches, one with POE injection, a few APs and a UniFi doorbell cam. The network is pretty solid, and I love the UniFi gear.

I have a tiny headless HP workstation sitting next to the NAS running Win10 with 4 cores and 16GB RAM, 256GB storage mostly idle. I’ve messed with using terraform to spin up VMs on Hyper-V with it, so I have some limited capacity there. I set it up to use an iSCSI connector with the NAS for a while when I had it running a Minecraft server, but always having a iSCSI connector attached to the NAS prevented the NAS from properly auto updating, and I’ve since removed that part of the setup there.

The mini cloud idea started because I want to play with Kubernetes and a better way to have some managed containers running all the time separate from my laptop and stuff. It would also be nice to experiment with some of the open source Serverless stuff to see what it takes to setup some of that. I want to use declarative tools such as Terraform, Packer, and Ansible to setup the VMs and then start experimenting with Kubernetes and Helm charts on top of those pieces. I want to play with Hashicorp Vault, Consul, and Nomad, and I want to setup a bunch of observability tools in an Elastic Stack. With those building block I could eventually build something/anything running locally as a distributed system, monitor and measure it all capturing distributed tracing and logging, then beat it up with some load testing tools or introduce some chaos to find where it all breaks. Lots of big ideas, and lots of work to build the foundation of a local playground for it.

I added a new Lenovo ThinkStation Tiny P-360 to the network last fall and slowly trying to figure out how I want to set it up to be able to provision VMs with Terraform on it. The new box that is about the size of an old external hard drive packs in a 12th gen Intel with 12 cores (20 threads / vCPUs) 32GB RAM, 1TB NVMe storage. I have it setup to operate headless wanting it to run as a Linux hypervisor. I had some expectations that I could just setup a simple Linux hypervisor and be going, but the journey hasn’t been so simple yet.

On the P360 I started trying with Ubuntu Server (because I plan on it being headless) but Server doesn’t have all the same hardware support as Ubuntu Desktop, and I had to figure out how to get all the Intel Pro 1000 drives files onto it to compile the network driver because what is shipped with the install media doesn’t work on the box. I then messed with getting Canonical MicroStack setup on Server but found it has some limitations for storage and networking with how I want to use Terraform to provision multiple VMs on it for use on my network. I’ve setup VMs with Terraform on other OpenStack cloud providers before (Rackspace Cloud, OVH) so I probably had too high of initial expectations for MicroStack.

While I got Ubuntu Server running eventually, I had some intermittent issues with basic networking that didn’t make much sense to me where it was like the network interface would just go to sleep after a while and not allow me to remote SSH into it, so I scrapped Server and setup Ubuntu Desktop, then figured out how to set it up so I can RDP into the GUI. After I got RDP working, I messed with installing the GPU driver for the Nvidia T400 4GB it has, so maybe I could mess with some of the AI things out there like using OpenAI whisper to translate and transcribe some video files.

The direction I think I’m headed is to get LVM and libvirt setup on the P360 running Ubuntu Desktop and target that with Terraform to provision VMs.