Diagnosing Cloud region selection on Build without repeated probes

We use the default global LiveKit Cloud URL with JS RTC 2.23.0 and Python RTC 1.1.21, with no application region hint, custom ICE, relay override or prepareConnection call.

A single standalone HTTPX GET to /settings/regions from an existing Render Singapore worker returned HTTP 200 with North American regions first; Singapore was absent. It used TLS verification, trust_env=False, no redirects or retries, and a short-lived token without room-join grants. A separate operator-side raw GET returned Hyderabad first, also without Singapore. These requests bypassed SDK region caches. The exact NAT source IPs were not captured.

Earlier closed WebRTC sessions showed India South in Cloud participant analytics. These sessions occurred before the raw GETs, so we are not treating the later region lists as proof of their original routing or fallback path. Historical native fallback logs are unavailable.

For a Build project using the default global URL, with no application region hint and no pinning control visible in the dashboard, what supported diagnostic fields or process can distinguish source-location classification from region eligibility/filtering and global-entry versus SDK fallback? Does the lookup token grant context affect the returned region list? How can the Cloud team correlate already-recorded requests and closed session IDs through a private channel, without publishing identifiers or logs here?

We want to establish the selection reason before changing anything. This is an information request, not authorization for a plan upgrade, pinning, proxy, recording or configuration change. No project domain, project/session/participant IDs, IPs, exact timestamps, credentials, audio, transcripts or personal details are included in this post.

I am not really understanding what you are asking.