ZeroGPU Space never gets scheduled ("Scheduling failure: unable to schedule") on new PRO account

#190
by ZAaraAa - opened

Hi Hugging Face team,

I subscribed to PRO on 27 Oct 2026 to host a Gradio Space on ZeroGPU, but the Space never gets a machine when ZeroGPU is selected.

Account: ZAaraAa (PRO, email verified)
Space: ZAaraAa/fanmeta-model-api (Gradio 6.29.1, Python 3.12)

What happens with hardware zero-a10g:

  • The build succeeds, then the Space stays at APP_STARTING with hardware.current = null.
  • My app never starts. The first line of app.py prints a message, and it never appears in the container logs (only "===== Application Startup =====").
  • After about 25 minutes the Space showed: "Scheduling failure: unable to schedule".

What I tried (all times UTC, 27 Oct 2026):

  • 02:35 and 03:31: deployed with zero-a10g. Stuck as described above.
  • 03:41: factory reboot. Same result.
  • 04:11: deleted the Space and created it again with the same name. Same result.
  • 04:25: switched the same Space to cpu-basic. It was scheduled within seconds and the app started immediately. It is now running fine on cpu-basic.

Your status page showed all services operational at that time. Because cpu-basic works, the problem seems specific to ZeroGPU scheduling for my account. Could you please check whether ZeroGPU hosting is fully enabled for my PRO account?

Thank you,
ZAaraAa

ZAaraAa changed discussion status to closed
ZAaraAa changed discussion status to open

Sign up or log in to comment