HTTP 429 Too Many Requests returned by /api/v2/bots/active/conversations during peak load simulation.
We are running a load test for a new AI-driven customer service bot in Genesys Cloud (v10.5) to determine the maximum concurrent session capacity. The test is executed using JMeter 5.6.2 with a ramp-up period of 60 seconds and a target of 500 concurrent users. The goal is to validate the bot’s response time and stability under stress. However, the test fails consistently after approximately 150 concurrent sessions are established. The JMeter logs show a sudden spike in 429 errors from the Genesys Cloud API endpoint /api/v2/bots/active/conversations. This endpoint is used to track active conversations for monitoring purposes in our custom dashboard. The error rate increases exponentially as the concurrent user count approaches 200. We have checked the API rate limits in the Genesys Cloud admin console, and the limit is set to 100 requests per second per user. The JMeter test is configured to respect this limit by using a Constant Throughput Timer. However, the 429 errors persist. We suspect that the WebSocket connection limit for bot sessions might be the bottleneck, but the documentation is unclear on the specific limits for AI bot conversations. The environment is US-East-1. The bot is configured with a simple NLP model and a basic response template. There are no complex integrations or external API calls involved in this test case. We need to understand the exact rate limits for bot conversation APIs and how to structure the JMeter test to avoid throttling while still achieving high concurrency. Any advice on configuring the JMeter test plan or adjusting the Genesys Cloud settings to handle higher concurrent bot sessions would be appreciated. We are also open to suggestions on alternative methods for monitoring active bot conversations that might be less resource-intensive.