Official Runpod ComfUI pod template fails to use my Network storage setup
The latest ComfyUI template's start.sh script forces a CUDA 12.4 to 12.8 migration even when persistent environments on /workspace are already compatible. This creates a loop where optimized user environments are renamed to .venv.bak and ignored, causing unnecessary 15-minute downtimes on every boot. The script should check for a 'force-skip' flag or verify if sqlalchemy/alembic are already satisfied before triggering a full re-install.
The auto-migration is ignoring my pre-configured setup on /workspace.
This issue has arisen since this morning. Yesterday, everything was running without issues!
The auto-migration is ignoring my pre-configured setup on /workspace.
This issue has arisen since this morning. Yesterday, everything was running without issues!
