comfyui, Runpod, 5090 + pytorch 2.8.0
Yesterday I created a 350GB disk, launched a pod with a 5090 + PyTorch 2.8.0, installed ComfyUI, configured everything step-by-step following the course I bought, and it worked perfectly — image generation, video generation from images, my custom LoRA.
I deleted the pod, and today I set up a new one.
When starting ComfyUI, I noticed in the logs: “pipe to stdout was broken”, but ComfyUI still launched normally.
When I tried generating an image, the logs showed: (screenshot) — it took half an hour to generate, and the next ones were just as slow.
Why are these errors happening? Yesterday the exact same configuration worked flawlessly.
GPU availability: medium. Pod restarted and recreated from scratch — no change.
I deleted the pod, and today I set up a new one.
When starting ComfyUI, I noticed in the logs: “pipe to stdout was broken”, but ComfyUI still launched normally.
When I tried generating an image, the logs showed: (screenshot) — it took half an hour to generate, and the next ones were just as slow.
Why are these errors happening? Yesterday the exact same configuration worked flawlessly.
GPU availability: medium. Pod restarted and recreated from scratch — no change.
