Keep blocking retries off shared media loops

Keep blocking retries off shared media loops

In a multi-session real-time media process, one synchronous database or HTTP lookup can pause every session when it runs on the event-loop thread. Retry backoff is especially dangerous because ordinary transient failures turn a short lookup into seconds of global audio starvation. Move blocking clients and sleeps to a bounded executor or use truly asynchronous clients, then make admission sensitive to event-loop lag rather than active-session count alone. This identifies an architectural hazard; profiling is still required to prove it is the dominant production cause.

Applies when: several real-time sessions share one asynchronous process. Keywords: audio gaps event loop blocking database retry concurrent calls