Can anyone explain why the Genesys Cloud Architect queue API returns 503 Service Unavailable errors when simulating high-volume inbound traffic? We are running a load test on the US1 environment using JMeter 5.6. The goal is to validate the maximum concurrent session limit for our inbound voice traffic before the upcoming holiday season.
We are targeting the /api/v2/architect/flows endpoint to dynamically update queue configurations based on real-time wait times. The flow logic involves a simple queue node with a target of 50 agents. When JMeter hits 300 concurrent calls, the API response time spikes to over 2 seconds. At 400 concurrent calls, we start seeing intermittent 503 errors.
The error response body is minimal:
{
"message": "Service temporarily unavailable",
"status": 503
}
We are using the Genesys Cloud SDK version 2.3.0 for authentication and initial setup. The JMeter script uses HTTP Request samplers with keep-alive connections enabled. We have verified that the network latency between our test server and the US1 edge is less than 50ms. The issue persists even when we reduce the concurrency to 250 calls but increase the duration of the test.
Is there a known rate limit or capacity threshold for the Architect API during peak load? We need to ensure that our dynamic queue configuration updates do not fail under high concurrency. Any guidance on best practices for load testing the Architect API or alternative methods to update queue settings in real-time would be appreciated. We are trying to avoid hardcoding queue parameters in the flow definition.