Just noticed that the screen recording retrieval endpoints start returning 503 Service Unavailable errors when we push our JMeter script to 300 concurrent agents performing screen captures. We are running Genesys Cloud in the US-East-1 region and using the standard REST API calls to initiate recordings via POST /api/v2/recordings/screen and then polling GET /api/v2/recordings/screen/{screenRecordingId} for status updates. The issue seems to trigger specifically when the polling frequency hits 10 requests per second per agent, which totals 3000 req/min across the load test. The error response body is minimal, just a generic service unavailable message with no specific retry-after header. We have verified that our API keys have the necessary recording:read and recording:write scopes enabled. The recordings themselves start successfully, but the metadata retrieval fails intermittently. Is there a known rate limit for screen recording API calls that differs from standard conversation APIs? We are trying to determine if we need to implement exponential backoff in our load test scripts or if this indicates a backend capacity issue for the recording service during peak concurrent loads.