I am facing a dilemma with my new server build. I have been collecting components to upgrade from my current HP Z420 Xeon E5-2690 V2 system, which has 128 GB of RAM. I have acquired a PSU, a Ryzen 9 3900X CPU, and an X470 Taichi motherboard that supports ECC. However, due to my lack of research and incompetence, I made a mistake in purchasing the wrong type of memory modules.
Initially, I bought LRDIMMs, but the system would not boot. Thinking that the LR (Load Reduced) technology might be the issue, I exchanged them for R-DIMMs. It was only later that I realized that the AM4 socket of my motherboard only supports unbuffered ECC RAM. Unfortunately, UDIMMs are more expensive than my budget for a 128 GB kit.
My main concern is that I am running a TrueNAS Core VM with an HBA passed through to it, and I have allocated 64 GB of RAM to it. The TrueNAS VM has a 50 TB storage capacity, and I would like to migrate it over to the new system. If I were to use non-ECC RAM, which I know is not recommended but less expensive, would it cause significant issues?
Alternatively, I am considering buying a single 32 GB stick of unbuffered ECC RAM and upgrading over time. However, I have three Ubuntu VMs running on my server, one of which is a Docker host with 20 containers, a Hassio VM, and a TrueNAS VM. Would 32 GB of RAM be sufficient to handle my needs? Kinda feel defeated been collecting parts for the past few months looking for deals and can’t believe I glossed over the fact AM4 only supports Udimms.