LiteLLM RPM Limit Enforcement Across Multiple Proxies with Shared Configurations

[dxu104]

I have two Docker containers running LiteLLM proxies on different ports (4000 and 4001) with the same RPM limit. When I send two requests in one minute to the same proxy, the RPM limit triggers. When I send one request to the first proxy and another request to the second proxy, the limit does not trigger. I want to ask if this is the expected behavior that each proxy enforces its own RPM limit, even though they share the same virtual key.

Redis is actively being used to manage the proxies.
And no data has been written to Redis.

I think my question is similar to this https://github.com/BerriAI/litellm/discussions/4063
and https://github.com/BerriAI/litellm/issues/6602 and https://github.com/BerriAI/litellm/issues/6938