I read the resource he posted – multipule times. I feel a bit dense as I am way over my head and often don’t know what I don’t know. Is there a forum that is more suitable for newbies and slow folk? I clearly am in that crowd.
That said, I appreciate all the help even if I can’t quite make sense of it – at least I know that I am missing something and keep digging at it.
Hell I even watched a few videos from the “Art of the Server” guy – though it seems that most of his content is 5-7 years old and the new stuff was about power consumption – which is not an issue that I have considered attacking.
I am a home user that is most interested in future proofing my server. I use it mostly for Plex and downloading media from newslists and torrents and for my personal cloud. I think I would like to try some vm stuff in the future.
I can’t follow the information that you are writing as I am not so sophisticated or knowledgeable. Are you saying that I should consider a stronger CPU?
No. Your CPU is perfectly fine.
Thank you @SmallBarky – I have been reading the stuff you send and its helpful. I am a bit old and slow – this post made perfect sense to me! That was a refreshing experience 
Not necessarily… We’ll try to be as helpful as possible, but vague requests can only result in answers that are either vague or off-the-mark (AI benefits here from being a natural hallucinator…).
You’re a home user with a good amount of storage (8 HDDs) and a usual assortment of light apps.
Does “Plex” includes transcoding, or not (or very rarely)? “Rarely” could be left to the CPU. “Often” or “required” would point to hardware transcoding: Typically an Intel CPU with an iGPU… but if you want to “future proof” for future codecs, you may want to have a dGPU (Arc A310), so you can upgrade the GPU later. In the latter case you could as well go for Ryzen/EPYC4000.
For the HDDs, ideally you’d go for a motherboard with 8 SATA ports and avoid adding a HBA.
Then comes the possibilities of VMs in the future…
VMs of what? Services such as HAOS and/or graphical VMs (Windows)? The latter would need its own GPU.
Is it running VMs in TrueNAS, or virtualising TrueNAS itself under Proxmox? In the latter case, you need to make sure you can isolate the chipset SATA controller for passthrough (yes, it gets technical…), or use HBA. With a HBA and a dGPU, plus a few NVMe drives for apps/VMs, you’ll need to make sure there are enough PCIe lanes, and that they can be distributed as required => dual x8/x8 PCIe slots rather than a single x16.
In the simpler scenario, the Pro WS W680M-ACE is suitable—though likely overkill (that’s the “futureproofing” part).
In the more complex scenario, the single x16 slot is a killer. You want x16/x0 + x8/x8 instead, for instance the full ATX W680-ACE. Or you go crazy with EPYC8000 “Siena” and a corresponding board: Plenty of PCIe slots for everything, plenty of RDIMM for VMs… but you’re over 150 W idle. And that’s a big scary server which will remain mostly unused/vastly underused if the “future” virtualisation project never gets up.
1 Like
I understood most of that! 
I think I am inclinded to go big or go home The more I read, the more I wonder if I should just go ahead and make this a Proxmox server and run Truenas in one vm and Plex in another. It makes total sense now (that I have watched 5+ videos) how I could use Proxmox and why I would then need the dedicated passthrough on the HBA. Eventually, I would move some of my crypto valadators to it and not need to be running those concurrently. (they are simple linux NUCs,) but no need tor them if the vms work well.
Additionally, I can serve out a vm for my wife so she can have a simple ‘all-in-one’ screen and still have access to better gaming if I install and passthrough another gpu … so that’s where my mind is going with this…
I see you’ve got several threads on the same topic (I responded in the other one 
). Given what you’re describing now I would recommend against ”go big” and trying to fit everything into one server. Consider a server which can reasonably fit your current identified needs, with the possibility to add more nodes in a Proxmox cluster later on. Ceph works extremely well for clustered data storage in this scenario once you’ve got 3+ nodes and at least 10Gb networking between. TrueNAS will be running in a VM on the node where you’ve got your spinning disks and can be used for ”slow” i.e. HDD storage, backups etc.
Edit: in that scenario, your first node should accommodate as a minimum: management and Proxmox cluster lan (typically the built-in NIC on motherboard), faster NIC for data traffic (10+ Gb NIC in PCIe), spinning disks and HBA in PCIe for NAS VM, boot disk (enterprise SATA SSD would be fine), room to add an enterprise NVMe disk (M.2 or PCIe slot) for clustered Ceph storage in the future. If you need SLOG or L2ARC in your NAS/ZFS you would do best with an M.2/PCIe slot for that also, but an enterprise SATA SSD off your HBA would work too albeit less performant.
1 Like
Your CPU choice seems over powered. If you aren’t using many Docker containers or VM’s, it’s doubtful you will ever need that much power.
You might like one of the new EPYC processors from AMD. The AM5 socket variety. Low wattage, but very good performance