Just noticed that the WebRTC media stream fails to establish for approximately 15% of sessions when JMeter pushes 500 concurrent softphone connections to Genesys Cloud US1. The test uses the standard WebRTC SDK v2.4.1 via a custom Node.js bridge script.
The SIP INVITE returns 200 OK, but the WebSocket connection drops shortly after with a 1006 error. No 429 rate limit errors are present in the logs, which suggests this is not an API throughput issue. The Architect flow is simple: Route to Queue → Agent. Queue capacity is set to 1000 agents.
Is there a known WebSocket connection limit per organization or per tenant that triggers under this specific load pattern? The latency to the edge is under 20ms. Need to understand if the drop is due to signaling timeout or media server resource exhaustion. Sharing the JMeter thread group config if helpful.