Expose configured public URLs in runtime observability

[josecelano]

Goal

Use the v3 public_url configuration values introduced by #1417 in health-check responses, metrics, and runtime logs without conflating them with a service's configured bind address or its post-bind ServiceBinding.

Background

Each service has three distinct concepts:

Concept	Source	Meaning
Configured bind address	bind_address configuration	The requested local socket bind target. It may be wildcard (0.0.0.0 or [::]) and may use port 0.
Service binding	ServiceBinding created after the socket binds	The protocol plus the actual local socket address. An OS-assigned ephemeral port replaces configured port 0, but a wildcard address remains wildcard. It is an identity, not necessarily a reachable URL.
Public URL	Optional v3 public_url configuration	The operator-declared external endpoint. It may differ completely from the bind address and service binding because of reverse proxies, NAT, TLS termination, or DNS.

internal_service_url is a possible future concept. It is not implemented by this issue and cannot be inferred reliably from a wildcard service binding.

#1417 stores and validates typed v3 public_url values but deliberately does not consume them at runtime. #1980 migrates runtime consumers to explicit v3 configuration imports. This issue follows both changes.

Scope

In scope

• Add an optional public-URL representation to health-check service details while preserving existing service_binding, binding, and service_type fields.
• Add an optional public_url label to relevant per-service metrics when configured.
• Add configured public_url, when present, to relevant service startup and request logs; retain service_binding as the local service identity.
• Define and test absent-value behavior: services without public_url remain valid and do not claim a public endpoint.
• Test that public_url, configured bind_address, and post-bind ServiceBinding remain distinguishable, including wildcard binding with an OS-assigned port.

Out of scope

• Changing how Include public service URL in configuration #1417 validates or stores v3 public_url values.
• Changing ServiceBinding or adding an internal_service_url type.
• Choosing a concrete reachable interface for wildcard listeners.
• Modifying the v2 configuration schema or supporting a v2 runtime fallback.
• Changing BitTorrent protocol behavior or URL path routing.

Compatibility decisions

Surface	Required behavior
Health check	Add an optional public_url field. Retain service_binding, binding, and service_type unchanged.
Metrics	Add public_url only when configured. Confirm and document the resulting Prometheus-series/cardinality effect.
Logs	Emit service_binding as the local identity and optional public_url as the operator-declared endpoint. Neither replaces the other.

Acceptance criteria

• A configured v3 public_url is exposed as an optional health-check field without replacing existing service-identity fields.
• Relevant per-service metrics expose public_url only when configured.
• Relevant startup and request logs identify the local service with service_binding and, independently, the configured public_url when present.
• A wildcard bind address with configured port 0 demonstrates three separate values: configured bind address, post-bind service binding, and configured public URL.
• Services without public_url preserve existing health-check, metric, and logging behavior.
• No internal_service_url implementation or torrust-net-primitives change is made.
• linter all and relevant tests pass.
• Manual verification evidence records both configured and absent public_url cases.

Verification plan

• Run linter all, focused tests for changed server, health-check, and metrics packages, and cargo test --workspace.
• Start a local v3 tracker with bind_address = "0.0.0.0:0" and public_url = "https://tracker.example.test/announce"; verify health checks distinguish public URL from post-bind wildcard service binding.
• Send an HTTP announce and query Prometheus metrics; verify configured public URL and existing server_binding_* labels are both present.
• Repeat without public_url; verify existing fields remain and no public URL is claimed or labelled.

Risks and trade-offs

• Public URLs can increase Prometheus time-series cardinality; restrict the label to configured per-service metric series and document the behavior.
• Health-check additions must be optional and additive.
• Logs and API fields must name service_binding and public_url explicitly so operators do not mistake either for an internal reachable URL.

References

• Include public service URL in configuration #1417 — typed v3 public URL configuration
• Make server listen URL available where server socket address is also available #1415 — service binding identity
• Configuration overhaul final cleanup: remove global re-exports, migrate consumers to explicit versioned imports #1980 — explicit v3 consumer migration
• EPIC Overhaul: Configuration (schema v3.0.0) #1978 — configuration overhaul