FreeSWITCH Health Is a Fleet Count, Not One Boolean

A multi-worker voice stack can keep serving calls after one FreeSWITCH node fails, so one global up/down flag hides degraded capacity. I ended up treating voip_freeswitch_nodes_healthy versus configured nodes as the useful observation point rather than relying on a generic service-up indicator. Comparing healthy and configured worker counts shows partial degradation before the entire service becomes unavailable.

This is a good example of redundant-capacity monitoring. The purpose is to reduce ambiguity during an incident: a signal should tell me which layer to inspect next, not simply confirm that something somewhere looks unusual.

For production I use the following guardrail: Alert when healthy workers fall below configured capacity, and keep per-node session and heartbeat panels for fast isolation. The same rule keeps the dashboard useful when the system grows and more targets are added.

The implementation is traceable to b65d5d4 in the hserver repository. That commit is the concrete reference for the collector, alert, dashboard, or runtime change behind this article.