Admin API

Snakeway includes a built-in administrative API for monitoring proxy health, inspecting upstream services, viewing traffic statistics, and triggering configuration reloads. All endpoints are served on the address configured in bind_admin under the /admin/ path, and all responses are JSON-formatted.

Configuration

The Admin API listens on the address specified by bind_admin in your server configuration:

server {
  bind_admin = "127.0.0.1:8081"
}

caution

The Admin API exposes operational data and a reload endpoint. Bind it to a loopback or internal address and restrict access at the network level.

Endpoint Reference

GET /admin/health

Returns the overall health status of the Snakeway instance and its registered upstream services, including per-upstream health checks and circuit breaker state.

curl http://localhost:8081/admin/health

Example response:

{
  "status": "healthy"
}

Error responses:

Status	Meaning
200 OK	Instance is healthy.
503 Service Unavailable	One or more critical upstreams are unhealthy.

GET /admin/upstreams

Provides a detailed view of all registered upstreams, including their current health status, circuit breaker state, and request counters.

curl http://localhost:8081/admin/upstreams

Example response:

{
  "services": {
    "api": {
      "127.0.0.1:8080": {
        "health": {
          "healthy": true
        },
        "circuit": "closed",
        "active_requests": 0,
        "total_requests": 150,
        "total_successes": 148,
        "total_failures": 2,
        "circuit_params": {
          "enabled": true,
          "failure_threshold": 5,
          "open_duration_ms": 10000,
          "half_open_max_requests": 1,
          "success_threshold": 2,
          "count_http_5xx_as_failure": true
        },
        "circuit_details": {
          "consecutive_failures": 0,
          "opened_at_rfc3339": null,
          "half_open_in_flight": 0,
          "half_open_successes": 0
        }
      }
    }
  }
}

The circuit field reflects the current circuit breaker state: "closed" (normal operation), "open" (upstream is failing, requests are rejected), or "half_open" (probe requests are being sent to test recovery).

GET /admin/stats

Returns aggregated traffic statistics per service, including active connections and cumulative counters.

curl http://localhost:8081/admin/stats

Example response:

{
  "api": {
    "active_requests": 5,
    "total_failures": 12,
    "total_requests": 1500,
    "total_successes": 1488
  },
  "static": {
    "active_requests": 0,
    "total_failures": 0,
    "total_requests": 450,
    "total_successes": 450
  }
}

Metrics per service:

Metric	Description
active_requests	Number of requests currently in flight.
total_requests	Cumulative requests handled since last restart or reload.
total_successes	Requests that resulted in a successful response (typically 2xx and 3xx).
total_failures	Requests that resulted in an error (4xx, 5xx, or connection failures).

POST /admin/reload

Triggers a hot reload of the Snakeway configuration. Snakeway validates the new configuration before applying it. If validation fails, the existing configuration remains active.

curl -X POST http://localhost:8081/admin/reload

Example response (success):

{
  "status": "ok",
  "epoch": 3
}

The epoch field is a version counter that increments with each successful reload.

Example response (validation failure):

{
  "status": "error",
  "message": "invalid configuration: unknown field `cidr_alow` in network_policy_device"
}

Error responses:

Status	Meaning
200 OK	Reload was successfully initiated.
400 Bad Request	Configuration validation failed. The response body contains the error message.

Circuit Breaker Transition Logs

When an upstream's circuit state changes, Snakeway emits a structured log event:

{
  "timestamp": "2026-03-21T15:27:00.000Z",
  "level": "INFO",
  "fields": {
    "event": "circuit_transition",
    "service": "api",
    "upstream": "UpstreamId(12345)",
    "from": "closed",
    "to": "open",
    "reason": "failure_threshold_exceeded",
    "failures": 5
  }
}

Transition reasons:

• failure_threshold_exceeded: Too many consecutive failures in Closed state triggered the circuit to open.
• cooldown_expired: The configured open duration elapsed, transitioning from Open to HalfOpen.
• half_open_failure: A failure during HalfOpen immediately re-opens the circuit.
• success_threshold_reached: Enough successful probes in HalfOpen state closed the circuit.

Using the Admin API for Monitoring

The /admin/stats and /admin/upstreams endpoints are designed to be polled by monitoring systems. A simple health check script might look like this:

# Check overall health
curl -sf http://localhost:8081/admin/health || echo "Snakeway is unhealthy"

# Fetch stats for a Prometheus exporter or custom dashboard
curl -s http://localhost:8081/admin/stats | jq .

# Inspect a specific service's upstream health
curl -s http://localhost:8081/admin/upstreams | jq '.services.api'

Stats counters reset on process restart or successful configuration reload. For long-term trend analysis, export the counters to an external time-series database at a regular polling interval.