I’m completely stumped as to why the Architect API is returning 429 Too Many Requests errors when we simulate moderate concurrent flow execution. The environment is US1. We are using JMeter 5.6 with the Genesys Cloud SDK 2.3.0. The goal is to validate the stability of our new IVR flow under load. The flow is simple. It has one menu prompt and two transfer actions. No complex logic. No external integrations. Just basic routing.
The test plan sends 200 concurrent requests per second to the /api/v2/architect/flows endpoint to fetch flow definitions. Then it triggers execution via the webchat channel. After about 10 seconds, the error rate spikes. The 429 responses are consistent. They do not stop. The retry logic in JMeter does not help. The headers show the retry-after time is very short, but the calls still fail.
Error 429: Too Many Requests. Rate limit exceeded for tenant. Please reduce request frequency.
We checked the API documentation. It says the default rate limit is 500 requests per minute for most endpoints. Our load is well below that. We are only hitting 1200 requests in the first minute. The error starts after the 500th request. It seems the limit is per second, not per minute. Or maybe it is per thread. The documentation is not clear on this.
We tried adding a delay between requests. That helps, but it reduces the throughput. We need to know the exact limit. Is it 500 per minute or 10 per second? The error message does not give enough detail. It does not specify which limit was hit. We are confused. The load test is blocked. We cannot proceed without knowing the correct threshold. Please advise. We need to adjust our JMeter config to stay within the limits. Or maybe we need to request a higher limit from support. What is the best practice? We are new to this. We want to do it right. The team is waiting for results. We are stuck. Any help is appreciated. We need to know the exact rate limit for the Architect API during high concurrency. Is there a way to see the current usage? The dashboard does not show real-time API call counts. Only historical data. That is not useful for debugging. We need real-time visibility. Please help.