Service "api" in project octagram-platform, environment production, region ams. Three separate deployments each got stuck indefinitely in DEPLOYING status, well past the documented 5-minute healthcheck retry window, never transitioning to SUCCESS or FAILED.

Verified facts:

• Container builds and starts cleanly every time: "Uvicorn running on http://0.0.0.0:8000", no crash, no restart, across all three deployments.
• The public service domain's target port (8000) matches exactly what the app listens on.
• Railway is NOT injecting a PORT environment variable for this service (confirmed via the API — no PORT present), ruling out a port-mismatch cause.
• curl -v against the public domain (api-production-579a.up.railway.app/health) shows a clean TLS/HTTP connection to Railway's edge (sin1), but returns 404 "Application not found" with header x-railway-fallback: true — i.e. the edge itself is reachable, but no deployment has ever actually gone live to route traffic to.
• HTTP proxy logs for all three deployments are completely empty (no probe traffic recorded at all), even during the healthcheck retry window.
• Tried: redeploying fresh (3x), and changing the healthcheck path — no change in behavior.

This looks like a stuck deployment / healthcheck prober issue on Railway's platform side rather than an application or config issue. Please investigate why deployments for this service/environment never leave DEPLOYING and never route traffic despite the container running healthily.