CRITICAL — Docker Space stuck in BUILDING state stalls: minutes and not resolving

Spaces
1

**
Priority: Critical** — This is blocking a production release cycle with a Tuesday deadline.

My Docker Space (SkipSnow/dev_ChatHealthyAIChatWindow) has been stuck in BUILDING stage for over 30 minutes following a successful git push. The Space is currently unresponsive.

This has been a recurring problem since we converted this Space to Docker SDK. Build times are unpredictable — sometimes 2 minutes, sometimes 30+ minutes with no progress.

We upgraded from cpu-basic to cpu-upgrade specifically to get better build times. That caused a different problem — a hardware mismatch where the Space requested cpu-upgrade but was allocated cpu-basic, leaving it stuck in BUILDING indefinitely. We had to factory-restart back to cpu-basic to recover from that incident.

The Dockerfile is a minimal 7-line file based on the HuggingFace Docker template — no custom build logic:

FROM python:3.12-slim
WORKDIR /app
COPY requirements.txt .
RUN pip install --no-cache-dir -r requirements.txt
COPY . .
EXPOSE 7860
CMD ["python", "main.py"]

Details:

  • Space: SkipSnow/dev_ChatHealthyAIChatWindow

  • SDK: docker

  • Hardware: cpu-basic (current and requested both cpu-basic, no mismatch)

  • Repo SHA: 54e2aa5329f0 (latest push, verified via /tree/main API)

  • Runtime SHA: f9335c700a9e (stale, from previous build)

  • Replicas: current=0, requested=1

What I’ve tried:

  • Factory restart (POST /restart?factory=true) — did not resolve

  • Regular restart — did not resolve

  • Build log API returns timeout (no log data streaming)

  • The previous container was killed by the factory restart, so the Space is now unresponsive

The code is confirmed in the HF repo. The Docker build queue appears stuck. Can someone from HF engineering check the build infrastructure for this Space?

1 Like