Stuck on a specific timeout issue when running load tests against a Genesys Cloud Architect flow. The setup involves a simple IVR with a single “Say” node followed by a “Hangup” node. Under low concurrency, the flow executes correctly. However, when increasing the thread count in JMeter to simulate 200 concurrent calls, the calls fail with a 504 Gateway Timeout after approximately 15 seconds.
Here are the steps to reproduce the issue:
- Create a basic Architect flow with a “Say” node (5-second audio) and a “Hangup” node.
- Configure JMeter with an HTTP Request sampler to POST to the
/api/v2/architect/flows/{flowId}/executeendpoint. - Set the thread group to 200 threads with a ramp-up period of 10 seconds.
- Run the test plan.
Observation: About 60% of the calls return a 504 error. The remaining calls complete successfully. The Genesys Cloud admin dashboard shows no increase in error rates or capacity issues during the test window. The API rate limits are not being hit, as verified by checking the X-RateLimit-Remaining headers.
The environment details are:
- Genesys Cloud Org: Production
- JMeter Version: 5.5
- SDK Version: v2.1.0
- Architect Flow ID: a1b2c3d4-e5f6-7890-abcd-ef1234567890
I have checked the network latency between the JMeter server and the Genesys Cloud endpoints, and it is stable at around 20ms. The issue seems to be related to how the Architect engine handles concurrent execution requests. Is there a known limitation on the number of simultaneous flow executions? Or is there a specific configuration in the Architect flow that needs to be adjusted for high-concurrency scenarios?
Any insights into this behavior would be greatly appreciated. I have attached the JMeter test plan XML file for reference.