Dealing with a very strange bug here with the predictive routing API when simulating high-volume inbound traffic.
Trying to validate queue capacity limits for a new campaign setup. Running JMeter from Singapore targeting Genesys Cloud v2 APIs. The goal is to simulate 500 concurrent call attempts to a specific predictive queue to see how the system handles overflow and rejection logic.
The test ramps up linearly over 60 seconds. Once concurrency hits around 150 users, the /api/v2/flexibleimaging/queues endpoint starts returning 503 Service Unavailable errors. This happens even though the actual queue depth in the admin UI shows plenty of available agents. The 503s are intermittent, not consistent, which makes it hard to pinpoint if this is a backend throttling issue or a misconfiguration in the JMeter script.
Environment details:
- Genesys Cloud Org: Sandbox (US-East)
- JMeter Version: 5.6.2
- Concurrent Users: 500 (ramp-up 60s)
- Target Endpoint:
/api/v2/flexibleimaging/queues/{queueId}/conversation - Error Rate: ~30% 503 errors after 150 concurrent requests
Has anyone seen this behavior with predictive routing endpoints specifically? I know standard APIs have rate limits, but I assumed queue interaction endpoints were handled differently for load testing purposes. Any tips on adjusting the JMeter configuration or understanding the specific limits for predictive queue interactions would be appreciated.