metrics

Prometheus Metrics Example

Instrument a go-mink event store with Prometheus metrics for observability.

Production event-sourcing systems need visibility into throughput, latency, and error rates. This example wires the go-mink metrics middleware into an event store, exercises it with simulated user traffic, and exposes a live Prometheus scrape endpoint so you can watch the counters and histograms fill up.

What this demonstrates

• Metrics instance — minkmetrics.New(...) builds a *Metrics configured with WithNamespace("mink_example") and WithMetricsServiceName("user-service"), which prefix and label every metric.
• Prometheus registration — MustRegister() registers the collectors with the default Prometheus registry so they can be gathered and scraped.
• Command middleware — CommandMiddleware() returns a mink.Middleware you plug into a command bus to record per-command counts, statuses, and durations.
• Live scrape endpoint — a background HTTP server serves promhttp.Handler() at :9090/metrics for Prometheus to poll.
• Simulated workload — appends registration, login, and profile-update events against the in-memory store to generate realistic metric activity.

Running

go run ./examples/metrics

No infrastructure required — uses the in-memory adapter.

This example starts an HTTP server on :9090/metrics and then runs until you press Ctrl+C.

What happens

1. Creates a memory.NewAdapter() event store and a metrics instance (namespace=mink_example, service=user-service), then calls MustRegister().
2. Launches the Prometheus HTTP server in a goroutine at http://localhost:9090/metrics.
3. Registers 5 users (user-001 … user-005) by appending a UserRegistered event per stream, printing each with its append latency in milliseconds.
4. Simulates 20 logins — loads a random user's stream, then appends a UserLoggedIn event.
5. Simulates 10 profile updates by appending ProfileUpdated events to random user streams.
6. Prints a metrics summary showing sample metric names, the final per-stream event counts, and the list of registered mink-prefixed metrics gathered from Prometheus.
7. Prints observability tips and keeps the /metrics endpoint alive until Ctrl+C.

Key APIs

• minkmetrics.New(opts ...MetricsOption) *Metrics — construct the metrics collector set.
• minkmetrics.WithNamespace(string) — set the metric name prefix (e.g. mink_example_...).
• minkmetrics.WithMetricsServiceName(string) — set the service label on all metrics.
• (*Metrics).MustRegister() — register the collectors with the default Prometheus registry.
• (*Metrics).CommandMiddleware() mink.Middleware — middleware to instrument command dispatch.
• mink.New(adapter) — create the event store over the memory adapter.
• store.Append(ctx, streamID, events, ...) / store.Load(ctx, streamID) — the operations being measured.
• promhttp.Handler() — Prometheus HTTP handler exposed at :9090/metrics.

Related

• Examples: tracing · full-ecommerce · basic
• Docs: Advanced Patterns · API reference