Stuck on 429 Too Many Requests errors when pushing 200 concurrent WebSocket connections to the analytics metrics endpoints. The rate limit header resets too slowly for our load test ramp-up.
The documentation actually says WebSocket rate limits are enforced per tenant, not per API key, so spreading load across keys won’t help.
Switch to batch analytics queries or increase the JMeter ramp-up significantly to stay under the 100 req/s threshold.
The documentation actually says WebSocket streaming is not intended for high-throughput load testing; use the batch metrics API instead. Switching to an analytics query allows you to aggregate data in chunks, bypassing the per-tenant WebSocket throttle entirely.
If I remember correctly… the batch endpoint is the only reliable path here, as detailed in Support Article 9942-B.