Observability Example

This example uses the in-memory noop broker, so it needs no Docker service or credentials. All new observability is off unless the master -observe switch is present.

go run ./examples/observability
go run ./examples/observability -observe -logs
go run ./examples/observability -observe -health -diagnostics
go run ./examples/observability -observe -metrics -correlation -events
go run ./examples/observability -observe -all -scenario=handler-failure

Switch	Purpose
-observe	Master opt-in switch
-all	Enable every category
-logs	Structured failure records
-health	Passive local health
-diagnostics	Safe diagnostic snapshot
-metrics	Bounded operation measurements
-correlation	Publish-to-handler context propagation
-events	Lifecycle and failure events
-probe	Explicit active probe; requires -health
-scenario	normal, connect-failure, publish-failure, or handler-failure
-timeout	Bound scenario and probe waiting; default 5s

Category flags without -observe are rejected. Diagnostic output never prints message bodies, raw connection strings, or correlation values. Synthetic failures are deterministic teaching aids and do not retry or alter acknowledgment behavior.