sed s/long-story/short/ I put together a bare metal VM host running Proxmox and the motherboard died just a few days into operation. I need a VM setup and I have a very large fresh-built headless NAS with lots of free RAM and CPU cycles. The OS is Ubuntu Server 16.04. In the meantime my plan is to run Proxmox using QEMU and KVM.
I've been reading through the Arch Wiki and man pages all morning. I know this is possible, but because of the sheer amount of options and power behind QEMU/KVM I'm having trouble sorting out what is the best way to execute the Proxmox guest.
The machine is a dual E5-2670v1 with about 100GB of RAM, OS running on a Sandisk flash drive plugged directly into the motherboard, with a SAS RAID 5 array powered by BTRFS. The 2670 supports VT-x, VT-d, and EPT so hardware support is 100%.
What are the specific QEMU options I should be aware of for this setup? I'm guessing just using conventional bridged networking to the Proxmox guest will allow Proxmox to provide bridged networking to the nested guests? I know in general the performance of the root file system for a VM host isn't going to be a big deal, but will it be any different on the nested VM setup? Will the overhead of using a qemu image instead of raw for the Proxmox root affect the performance of the nested VMs? What's the most lightweight no-fuss way to get access to the Proxmox installer? It's all web-based after install so I only need graphical access to the Proxmox guest once to get it installed and make sure networking is functioning for the web UI. Is there a way to just assign the Proxmox guest direct access to one physical CPU?
Sorry for the question bukakke. There's just so much to QEMU and I've only ever used it with graphical front-ends.
[link] [comments]