I’m trying to figure out why the Genesys Cloud platform_api WebSocket connections drop unexpectedly when load testing from Singapore. The setup uses JMeter 5.6 with a custom WebSocket sampler plugin. Target region is US-East, but the load generator is in Asia/Singapore to simulate real-world latency for APAC users.
Config details:
- Threads: 8,000 concurrent
- Ramp-up: 30 seconds
- Loop count: Forever (duration 15 mins)
- Auth: OAuth token refreshed every 50 minutes via HTTP Request
Everything works fine up to 5,000 threads. Once crossing that threshold, we start seeing connection resets. The logs show:
Error: WebSocket closed abnormally with code 1006 (Abnormal Closure)
The server-side logs in GC don’t show any corresponding errors. Just silent drops. Is there a hidden connection limit per organization or per API key for WebSocket endpoints? The docs mention rate limits for REST calls, but WebSocket persistence limits are vague.
Also, is the latency from SG to US-East causing the handshake timeout before the load completes? JMeter shows the average response time jumping to 4s right before the drops. Any tuning tips for JMeter WebSocket keep-alive settings in this scenario?