so i plan on having unraid in a proxmox vm and i have some questions.
the reason i’m doing this is because i want more control of vms and docker + i’ll be diving into kubernetes and i don’t want to use unraid baremetal and then be limited in what i do in anyway. basically unraid is going to be for storage and plex related containers ( sonarr, radarr, plex, etc…) and everything else would be on proxmox.
I don’t use unraid at the moment but if you aren’t going to use it for the docker and other functions then I wouldn’t bother using it at all. One benefit of unraid is the community support for the other systems around the harddrive space otherwise you anything else. Based on that I would use truenas core and ZFS for the share drive and nas system - and it’s free. Then as you want to run proxmox and do all your docker- vm-etc. Do that -VM one 2 core - and say 16gb of ram to a truenas core VM. Then leave the rest of your stuff for docker containers right on proxmox or whathave you. I do this now by the way.
Drives I wouldn’t setup that way. The spinners. I would just run one consistent pool and with the new ZFS updates if you have to add to them, you can. For example a 4-4tb setup in Z1 would give you about 11+Tb of useable. and you can tollerate one drive failure and if it does, drop in your replacement and the system will resliver. Then on proxmox for the otherside do whatever you want to do and point over to your shares.
Another options is to let proxmox run your zfs pool - something I see version 7 will do. I don’t know much about it but i thought about trying that out.
OH and what I do is I have a sdd that is Proxmox and trunas and the image repositiory, and another ssd that is all the VM and continer work. so they are seperate but it doesn’t have to be. my drives were only 128 and 256 gb large. so the 128 is the first one and the 256 is the VM and other stuff with lots of room. the docker that runs the RR’s uses 48gb with 2 core and 16 gb. the VM that runs linux with jellyfin and home assistant runs 4 core and 31gb and has 100 gb of drive (I plan to add to it)
SO yes I wouldn’t recommend doing what you’re planning but I offered an alternative.
why not? wouldn’t i be getting the best of both worlds if i run unraid in a vm ? all i need to do is passthrough the gpu, hba controller, nvme and the ssd for cache. on the proxmox side there will be 2 ssds one for proxmox itself and one for docker and vms. i think there is nothing wrong with this. i might just be spewing nonsense but that the reason i’m asking since i don’t know if what i’m saying is feasible or my thoughts are way off rails
There’s plenty wrong with it.
Unraid itself is a hypervisor and a docker container host. Why involve proxmox?
Passing hardware to Unraid is going to cause you a whole slew of headaches.
It’s certainly possible to do this, but if your goal is storage, Plex, and learning k8s, you could also flip it around and run Unraid baremetal and run a debian VM (or several) for a k8s node. Unraid can run the media acquisition dockers just fine. Add cheap 7th-gen uSFF or laptop for QSV Plex transcoding. For playing around with k8s, kind and k3d are great options with low barrier to entry.
I’m wasting all types of money setting up high speed and low speed, high core and low core Proxmox nodes and screwing around with them.
One thing I have no intention to debug is hardware pass through or optimization of unraid in a VM.
It’s bad enough I waste so much time fixing the Mellanox nic’s that cause MAC dupes (it seems Unraid finally fixed this), can’t imagine how annoying it would be within a vm.
It might cost a little more money, but get some various TMM machines to learn k8s.
In the old days I used to run unRaid under VMWare as I was a VMWare Admin for work. It was a pain. As unRaid grew up and gained new abilities I went bare metal unRaid and never looked back. With unRaid’s Dockers and VM capabilities it really is a one stop shop for me and has been rock solid for the 10 years I have been using it.