Trying to understand the connectivity behavior when deploying a custom analytics module via our AppFoundry integration into a Bring Your Own Cloud (BYOC) environment. The application is designed to aggregate real-time interaction data from multiple Genesys Cloud organizations and push it to a third-party data warehouse. While the REST API endpoints function correctly for historical data retrieval, the WebSocket connection required for live streaming events fails consistently.
The deployment uses standard OAuth 2.0 client credentials flow for authentication, and the access token is valid. The issue arises specifically when attempting to establish the WebSocket handshake at wss://api.mypurecloud.com/api/v2/analytics/events. The connection drops immediately after the initial handshake attempt without returning a specific error payload in the response body, only closing with a generic status.
WebSocket connection failed: 503 Service Unavailable
Handshake request: GET /api/v2/analytics/events?access_token=…&event_types=interaction-event
Response headers: Connection: close
This behavior is consistent across three different BYOC deployments located in the us-east-1 region. Our internal logs show that the request is reaching the Edge nodes, but the subsequent routing to the analytics service appears to be rejecting the connection. We have verified that the client IP addresses are whitelisted in the BYOC security settings and that the firewall rules allow outbound traffic on port 443.
Given that the REST API works, we suspect this might be related to how the Edge load balancer handles WebSocket upgrades in a BYOC context versus the standard SaaS environment. Are there specific limitations or known issues with WebSocket connections through the BYOC Edge nodes? We are currently using the latest version of the Genesys Cloud SDK for Node.js, but we have also tested with raw HTTP requests to rule out SDK-specific bugs.
Any insights into the expected behavior or troubleshooting steps for this specific 503 error in a BYOC setup would be greatly appreciated. We need to ensure stable real-time data ingestion for our partner clients.