I’ve spent hours trying to figure out why our Genesys Cloud BYOC trunk is dropping connections with a 408 Request Timeout when we push the concurrency past 200 calls per second in JMeter. We are using a simple loop controller to generate SIP INVITEs to a specific extension group. The initial burst of 100 calls connects without issue, but as soon as the thread group ramps up, the 408 errors start flooding the logs.
Our edge logs show the SIP messages are reaching the Genesys Cloud edge, but the response never comes back. We have checked the JMeter HTTP Request Defaults and ensured the keep-alive is disabled to prevent connection pool exhaustion on the client side. The network latency between our load generator in Singapore and the closest Genesys Cloud region is under 50ms, so that shouldn’t be the bottleneck.
Is there a specific rate limit on the SIP signaling path for BYOC trunks that isn’t documented? Or are we hitting a WebSocket connection limit on the edge before the SIP stack even processes the INVITE? Any insights on how to tune the JMeter timer or the SIP protocol controller to avoid this timeout would be appreciated. We are currently using JMeter 5.6.2 with the SIP plugin.