Quick question about handling 503 Service Unavailable when simulating high concurrency in JMeter.
Environment: Genesys Cloud US1
Tool: JMeter 5.6
Endpoint: /api/v2/recordings/search
We are running a load test to validate the capacity of the Screen Recording API. The goal is to simulate 500 concurrent agents requesting recording metadata simultaneously. The test starts with a login sequence to obtain valid access tokens. Then it moves to the recording search endpoint.
At low concurrency, say 50 threads, the API returns 200 OK within acceptable latency. The response time is around 200ms. However, when the thread count increases to 200, we start seeing intermittent 503 errors. At 500 threads, more than 60% of the requests fail with 503 Service Unavailable. The error message in the response body indicates that the backend service is temporarily unable to handle the request.
We have checked the API rate limits documentation. The documented limit for this endpoint seems to be sufficient for our test volume. We are not hitting the 429 Too Many Requests error, which suggests the issue is not a rate limit violation. Instead, it appears to be a server-side capacity issue or a timeout problem.
Is there a known limit on the number of concurrent connections for the Screen Recording API? Are there any best practices for handling these types of load tests? Should we implement exponential backoff in our JMeter script, or is this a sign of a deeper infrastructure issue? Any insights from the community would be appreciated. We want to ensure our reporting tools can handle peak loads without failing.