Windows 11 WSL Fixes RAM/ComfyUI

[BigRaver]

The Core Problem
HAMI (Olares' GPU virtualization layer) loads itself into every process via /etc/ld.so.preload. In WSL2, HAMI's NVML queries return -1 which causes it to panic and block CUDA entirely.
Fix 1 — WSL2 RAM (do this right after install)
Olares installer overwrites .wslconfig to 12GB. After install restore it, Find .wslconfig and overwrite with the following if you have the same set-up(Setup below):

[wsl2]
memory=52GB
processors=16
swap=13GB

[experimental]
autoMemoryReclaim=gradual

Fix 2 — DATA_DIR env var
run this in the wsl root:
kubectl patch deployment comfyuishare -n comfyuisharev2server-shared --type=json -p='[{"op": "add", "path": "/spec/template/spec/containers/0/env", "value": [{"name": "DATA_DIR", "value": "/root"}]}]'

Fix 3 — Disable HAMI via empty preload mount (the key fix)
run this in the wsl root:
kubectl create configmap empty-ld-preload -n comfyuisharev2server-shared --from-literal=ld.so.preload=""

kubectl patch deployment comfyuishare -n comfyuisharev2server-shared --type=json -p='[
{"op":"add","path":"/spec/template/spec/volumes/-","value":{"name":"empty-ld-preload","configMap":{"name":"empty-ld-preload"}}},
{"op":"add","path":"/spec/template/spec/containers/0/volumeMounts/-","value":{"name":"empty-ld-preload","mountPath":"/etc/ld.so.preload","subPath":"ld.so.preload"}}
]'

This is how i got ComfyUI running on Olares. i kept getting an error and finally figured it out and got it working properly! had major issues with it trying to find my gpu. Had issues with RAM in Olares but fixed that right away
My set up is:
RTX 4090 24GB VRAM
64gb RAM
i9-10850k
Windows 11 Home
Going to keep exploring and try to get everything to work and find any more bugs!