need some help troubleshooting
running a load test on agent scripting endpoints. the goal is to validate api throughput for dynamic script updates under high concurrent call volumes. using jmeter 5.6 on genesys cloud us1.
the test hits 50 concurrent threads against /api/v2/agent-scripts and immediately returns 429 too many requests. this happens even with a simple get request to list scripts. the error response includes:
{
“code”: “too_many_requests”,
“message”: “rate limit exceeded”,
“status”: 429
}
we are testing the capacity for updating script variables during active calls. the current setup uses a single api key for all threads. the documentation mentions rate limits for agent scripting apis but does not specify the exact threshold for concurrent requests.
here is the jmeter configuration:
- thread group: 50 threads
- ramp-up: 0 seconds
- loop count: 1
- request method: get
- endpoint: /api/v2/agent-scripts
- headers: authorization bearer token
the test fails immediately. no successful responses. this suggests the rate limit is very low or there is a misconfiguration in the api setup. we need to determine the correct way to configure rate limiting for agent scripting apis when running high-concurrency load tests.
is there a specific header or parameter to increase the rate limit? or is this a hard limit for the us1 environment? any guidance on how to structure the test to avoid hitting this limit would be appreciated. we are trying to validate the system’s ability to handle dynamic script changes at scale.
the current error pattern is consistent across multiple runs. the 429 responses are immediate. no delay or retry logic in the jmeter script. the api key has full admin permissions. the issue seems to be related to the number of concurrent requests rather than the frequency.
please advise on the best practice for testing agent scripting api throughput. we want to ensure the system can handle the expected load during peak hours. any insights into the rate limit behavior for these endpoints would be helpful. thanks.