Agent connect fails: "failed to retrieve region info" on LiveKit Cloud's own settings/regions endpoint

Agents
,
1

Agent connect fails: “failed to retrieve region info” on LiveKit Cloud’s own settings/regions endpoint

My agent (CA_mrcHWXWq7HKo, deployed on LiveKit Cloud) receives job requests correctly every time a room is created, but fails during connection with:

ConnectError: engine: signal failure: failed to retrieve region info: error sending request for url (https://grwth-coach-lel1ks86.oashburn1b.production.livekit.cloud/settings/regions)

This is a LiveKit-hosted endpoint being called by a LiveKit-hosted agent — there’s no client config or network setting on my end involved in this specific request.

Failed identically across three separate job attempts:

  • AJ_JfYWa2UtkXPG
  • AJ_XV3L5w4nd6A8
  • AJ_iemKydF6ufSK

Support’s initial response was that this isn’t a wider outage, but the failure is clearly internal to LiveKit’s own infrastructure (one LiveKit service failing to reach another). This has completely blocked my production voice agent since deployment. Has anyone else hit this specific settings/regions failure, or does anyone from LiveKit have insight into what’s going on with that endpoint?

2

@Logan_Knecht, This specific error is very likely not a LiveKit infra failure. “failed to retrieve region info” is a known and misleading symptom of the agent container missing the system CA certificate bundle: the SDK’s HTTPS call to /settings/regions fails TLS validation, and it surfaces as a region-info error rather than a certificate error. LiveKit documents this directly, since the ca-certificates line in the Node Dockerfile template is load-bearing because “node:22-slim doesn’t ship the system CA bundle, so Node agents on it fail with a misleading failed to retrieve region info error” (agents-js#1622, livekit-cli#849). Your exact signature, error sending request for url (…/settings/regions), is also tracked on agents-js#932.

The fix is to install CA certificates in the agent image. On a Debian slim base:

RUN apt-get update && apt-get install -y --no-install-recommends ca-certificates \
    && rm -rf /var/lib/apt/lists/*

(Alpine: apk add --no-cache ca-certificates.) Job requests arriving while the room connect fails still fits this: dispatch is delivered over the worker’s existing connection, while each room connect makes a fresh HTTPS request to /settings/regions, which is the call hitting the missing cert store. If you have confirmed the CA bundle is already present in your deployed image and still see it, that is the point to push back to support with your Dockerfile, but in the large majority of these reports it is the missing certificates.

3

Thanks Muhammad!

@Logan_Knecht Muhammad’s response fix your issue? I see agent sessions again listed in your dashboard, so I believe you are now up and running again. I will close the separate ticket you raised and handle responses in this thread.