No idea why this is happening, the web rtc softphone sessions are dropping with a 408 request timeout error when we push concurrent users above 500. we are using the genisys cloud web sdk version 2.0.150 in our jmeter script. the initial handshake works fine, but once the media stream tries to establish, the websocket connection times out.
our jmeter config uses a constant throughput timer to simulate realistic call patterns. the environment is a standard public cloud deployment. we are hitting the /api/v2/communications/conversations endpoint to initiate the calls. the error log shows:
websocket error: connection closed abnormally [code=1006, reason=""]
followed by
http 408: request timeout on /api/v2/communications/conversations/{id}/answer
we have checked the api rate limits and they seem to be within bounds for our org tier. is there a specific websocket limit or server-side timeout we are hitting? the latency from our singapore based load generators to the us-east-1 region is around 180ms. anyone else see this with high concurrency web rtc tests?