Dell T430, and 2x T620 options plus software stack options

I am in the very fortunate situation where I have been given 3 different Dell servers from work that have all reached end of warranty so have been replaced.

Server 1:

Dell PowerEdge T430

2x Intel Xeon E5-2630 v3 2.4GHz (2x 8C/16T))

64GB DDR4 ECC (8x 8GB)

2x 2TB 10K SAS HDD in RAID1 connected to a PERC H730)

8x3.5” HDD chassis

Server 2:

PowerEdge T630

1x Intel Xeon E5-2620 v3 2.4GHz (6C/12T)

32GB DDR4 ECC (2x 16GB)

8x2TB 7.2K SATA HDD connected to PERC H730P

8x3.5” HDD chassis

Server 3:

PowerEdge T630

2x Intel Xeon E5-2620 v4 2.1GHz (2x 8C/16T)

96GB DDR4 ECC (6x 16GB)

7x 600GB 10K SAS HDDs + 3x 1.2TB 10K SAS HDDs

16x2.5” HDD chassis

Old Desktop PC:

Intel i5-3570, 16GB DDR3, 2TB Seagate Hybrid SSHD

Now I am currently running:

Server 1 with Proxmox on it, currently only running UniFi Controller and 2 instances of PiHole (as LXC containers) and HassOS (as VM)

Server 2 with a 120GB Kingston SSD for OS running TrueNAS Core

Old Desktop PC running Win10 purely for BlueIris

What I am wanting to do is expand on my self-hosted apps, including: Plex, Grafana, Paperless-NGX, Radarr, Sonarr, Overseerr, Calibre, Bitwarden, Heimdall, potentially Nginx or Traefik, Tandoor Recipes, LibrePhotos, PlexPy,

I want to use Docker (with Docker Compose) and also Ansible (along with Ansible Vault).

Hardware wise, what I am wondering is, while these servers are seriously overpowered for what I am running, I think it would take a while to save the difference in power consumption vs the cost of buying/building a “new” PC with a desktop grade CPU, so I am thinking I’ll stick with the hardware I have. What I am wondering is for Plex with hardware transcoding, would I be better off to add a GPU to the server to offload some of the load from the CPUs, and maybe make it handle more simultaneous transcoding streams? I know the iGPU in the Desktop CPUs is the most efficient way to do it but that will cost more money and I already have this hardware…

Software wise, what I am wondering is, is Proxmox still the best option if I want to run most of the apps in Docker, or should I be switching to something like Ubuntu and running Docker straight on top of that. I’m thinking if I go with Proxmox, I’d end up with just a VM with Ubuntu on it and then it would have Docker on it. While I like Proxmox, I am not sure if its adding an extra layer of complexity that isn’t required. I know you can install Docker on top of Proxmox but I wonder if that’s a bit risky as you moving away from the “managed” nature of Proxmox and an update could break it.

I am happy to start from scratch with my software stack as I really want to use Ansible for the whole setup so that I can have all the playbooks backed up to GitHub and my data backed up as well, and then if the server was to completely blow up or be stolen or something else that required new hardware I can just re-run the playbooks and have restore the data and be back up and running quickly rather than having to re-configure everything from scratch.

I’m also wondering if its best to seperate the storage on to the NAS or if I save a bit more power and put extra HDDs in the main server?

I am considering potentially buying a cheap Dell/HP/Lenovo SFF PC with an 8th get Intel for faster Blue Iris performance, not sure if I should go with a USFF with a T series CPU or just go with a SFF.

I played with a few setups for my server and at this point I’m running trueNAS Scale release on mine. But I also tested and ran for a while a proxmox setup. And there I did 2 VM’s One truenas core for NAS work alone. So zfs pool and setup, samba share and I think I also had a NFS share and that’s it.

Then VM2 was different flavors of Ubuntu - started with Server 22.04 and off that I ran Docker and used docks for the ARR’s and for Jellyfin which was my media app. and it was cumbersome to me in that there was too many extra things. but it worked ok other than transcoding - that seems to screw the pooch hard. and that’s software transcode not hardware and with the existing server machine.

Ironically with Truenas Scale and the ARRS and Jellyfin - transcoding does work without stopping up the works. But I also think I have a flawed video. so meh, could have been either.

My bigger issue was all the things that ran. To work with and plan NAS go to this address - to play Docker go to this or this or this . . . . . . I was on the way of putting overseer on there when I decided to ditch that approach.

SO I’m using truenas scale at the moment. I also toyed with trying Homelab OS which I read a good things about has some of what you are looking to. and I thought about just running linux on the VM and no docks for any of the applications. I’m still uncertain as to why that wouldn’t work well if that VM instance was ONLY FOR search and downloads.

Also for my VM mix test I was giving 2 core and 16gb of ram to Truenas Core - and giving 6 cores and 48gb to Ubuntu. I have a similar processor to your one server with 8 core 16 thread it’s a 2590 or something like that. Also because I got tired of the text only life of ubuntu server I put on a flavor of Kubuntu that is the lighter of the family. I know of a guy that uses fedora something for his rig.