Could someone clarify why the Platform API is returning 429 Too Many Requests when I am trying to simulate a moderate load on our Architect IVR flows? We are running a basic JMeter test from our Singapore office, targeting the api/v2/architect/flows/{flowId}/sessions endpoint to initiate concurrent voice interactions. The goal is to validate how the system handles sudden spikes in inbound call volume during peak hours in the APAC region. We are using the standard OAuth token for authentication and have verified that our token refresh logic is correct. However, even with just 100 virtual users hitting the endpoint with a 2-second ramp-up time, we start seeing rate limit errors immediately. The JMeter log shows a mix of 200 OK responses and 429s, but the 429s appear almost randomly, not just after a specific threshold. This makes it difficult to determine the true capacity of the IVR logic. We are not making any other API calls during this test, so it is not a global rate limit issue across different endpoints. The error response body indicates that the rate limit has been exceeded, but given the low concurrency, this seems premature. We are on Genesys Cloud version 24.2.0 and using the latest version of the JMeter HTTP Request sampler. Is there a specific rate limit for session creation that is lower than the general API limit? Or is there a recommended way to stagger these requests to avoid triggering the rate limiter? We need to understand the baseline capacity before we move to higher volumes. Any insights into how the rate limiting algorithm works for this specific endpoint would be appreciated. We have checked the documentation, but it does not provide clear numbers for session creation limits.
“Rate limits are applied per organization and per endpoint. Exceeding the limit results in a 429 status code. Implement exponential backoff to handle retries.”
This advice is clear, but we are not exceeding the documented limits, at least not according to the public docs. We are confused about the exact threshold. Has anyone else faced this issue with IVR session creation? What is the typical throughput we should expect?