Observability

You cannot improve what you cannot see. When your applications depend on remote APIs, the traffic between them is often a black box, hard to inspect when something slows down or breaks. HARP sits on that boundary as a proxy, so every exchange flows through one place that can record, display and measure it. Without changing your application code, it gives you a dashboard to browse traffic, a metrics endpoint to feed your monitoring stack, and an audit log of what was sent and received.

Understanding the use case

Observability is the ability to understand the internal state of a system from the data it emits. For applications that call external APIs, the most valuable signals live at the edge: which requests were made, how the remote services responded, how long they took, and which ones failed. Because HARP forwards this traffic, it is naturally positioned to capture those signals for every API Transaction without instrumenting each application by hand.

This matters both during development, when you need to see exactly what a third-party API returned, and in production, where you need to spot degradations early and keep a record of past exchanges for later analysis.

Challenges and solutions

• Inspecting traffic to remote APIs: When an integration misbehaves, you need to see the actual request and response, not a reconstruction. The HARP dashboard, available by default on port 4080, lists every transaction that passed through the proxy and lets you open any one of them to read the full request and response headers and body. Filtering helps you narrow a busy list down to the exchanges you care about.

• Keeping a usable history: A trace is only useful if it is still there when you go looking for it. HARP retains transactions for a configurable period so you can review past exchanges, and prunes older ones automatically to keep the store bounded.

• Feeding your monitoring stack: Dashboards are for humans, but alerting and trends need machine-readable metrics. HARP exposes Prometheus metrics at /.prometheus/metrics, enabled with harp-proxy server --enable metrics. The metrics cover request and response counts, durations and in-progress gauges at the ASGI, controller and HTTP client levels, so you can chart latency, throughput and error rates for the traffic HARP handles and wire them into your existing alerts.

• Maintaining an audit log: Beyond live debugging, you often need a durable record of what was exchanged with which API. HARP keeps an audit log of API exchanges, giving you a reference you can return to when you need to reconstruct what happened.