Skip to content

TROUBLESHOOTING.md

Latest commit

 

History

History
92 lines (64 loc) · 3.67 KB

TROUBLESHOOTING.md

File metadata and controls

92 lines (64 loc) · 3.67 KB
Raw

Troubleshooting

start.bat says "Python not found"

Python is missing, or not on PATH.

  1. Reinstall from https://www.python.org/downloads/
  2. Tick "Add Python to PATH" on the first installer screen
  3. Open a fresh Command Prompt: python --version must print 3.10 or higher

The GUI window never appears

start.bat launches Edge (or Chrome) with --app against a loopback server. Run python moon_bridge.py --serve, open the printed URL in any Chromium browser, and read the console: a missing dependency or a refused port shows up there.

The server exits by itself after 12 s with no requests — the page polls every 80 ms, so "no requests" means "the window is closed". If your browser was still opening slowly, just run it again.

fuckingfast.co returns 403 on every link

curl_cffi is missing. Cloudflare fingerprints the TLS ClientHello, and aiohttp's scores as a bot no matter which headers it sends.

pip install curl_cffi

The engine prints a warning at startup when the import fails.

datanodes.to: "Verification failed" / Error 600010 in the Turnstile widget

Turnstile is refusing the browser, not the account.

  • Make sure a real Chrome or Edge is installed. Playwright's Chromium is not a Google-branded build and Turnstile rejects it. Set the path explicitly in the GUI, or MOON_CHROME_PATH=C:\Program Files\Google\Chrome\Application\chrome.exe.
  • Do not force headless. MOON_DN_HEADLESS=1 breaks Turnstile by design — the challenge platform answers 401 and the token never arrives.
  • Lower Pages to 1–2. Many tabs solving challenges at once from one IP looks like a farm.
  • Solve one manually if it asks: the clearance lands in the persistent profile and the rest of the session reuses it.

Chrome opens even though I only pasted fuckingfast links

Fixed in V2. moon_extract.BrowserGate launches on the first datanodes link and never before, in the WebView GUI, the Tk GUI and the CLI alike.

If you still see it, you are running pre-V2 files: check that moon_extract.py contains class BrowserGate and that moon_engine.py / moon_cli.py contain no bare open_browser( call. pytest tests/ -q answers this in a second.

Chrome will not attach: "could not start chrome.exe on port 9222"

Chrome refuses --remote-debugging-port on a --user-data-dir that another Chrome process already has open.

  • The app uses its own profile (%LOCALAPPDATA%\MoonDownloader\chrome-profile), so this usually means a leftover MoonDownloader Chrome. Close it, or kill the process.
  • Port 9222 taken by something else: MOON_CDP_PORT=9333.
  • Never point MOON_CHROME_PROFILE at your daily-driver profile — you get a browser you cannot attach to.

Downloads stall at low speed and get killed

Two different things look identical:

  1. A slow datanodes lane. The CDN pins your session to a lane; some are slow. The stall killer re-extracts to get a new one, up to STALL_MAX_KILL times.
  2. A saturated uplink. If the pipe is already full, every stream looks stalled. Lower DL streams.

Extraction dies partway through a long session

Fixed in 14.6. The shared Chrome used to die after ~80 sequential extractions and every later link failed with "Target page, context or browser has been closed". ensure_live_browser() now re-validates and respawns it transparently. If you see that message again, attach moontech_*.log to a bug report.

Where are my logs?

Next to the scripts:

  • moontech_YYYYMMDD_HHMMSS.log — human-readable
  • moontech_YYYYMMDD_HHMMSS.json — per-file metrics
  • CLI runs write moontech_cli_*.log / .json

Attach the .txt (or .log) to any bug report. MOON_DEBUG=1 adds extraction-level tracing.