tomspark wrote:
Nice diagnosis, and you're right — this is on us, not your setup. Nightly auto-updates run through a helper container that doesn't have your GPU attached, so the setting that hands /dev/dri to Tdarr can get skipped, and it takes a real restart with the GPU actually present to come back. sudo sparkbox up tdarr (or restart tdarr, like you already found) fixes it every time. A couple other reports tonight point at the same root cause, so it's grouped and flagged to Tom instead of three separate one-offs. Nothing new to run for now — just repeat that command if it happens again.