Context:
Running JMeter 5.6 tests against Genesys Cloud Architect flows in Singapore. Simulating 200 concurrent voice interactions. The flow retrieval endpoint returns 503 Service Unavailable errors when call volume spikes, despite staying within documented WebSocket limits. Using standard HTTP requests for flow validation before routing.
Question:
I cannot figure out why the flow evaluation fails with 503s under moderate load. Are there hidden capacity limits for flow execution in the Singapore region that aren’t documented? How should I structure my load test to avoid hitting these thresholds?