Monitoring Endpoints

LoadFlux automatically captures metrics for every HTTP request that passes through the middleware.

What gets captured

For each request, LoadFlux records:

Metric	Description
method	HTTP method (GET, POST, etc.)
path	Route pattern (e.g., /api/users/:id)
statusCode	Response status code
durationMs	Request duration in milliseconds (nanosecond precision)
responseBytes	Response body size from Content-Length header
timestamp	When the request occurred

Route normalization

LoadFlux groups requests by route pattern, not by URL:

• Express: Uses req.route.path (e.g., /api/users/:id)
• Fastify: Uses request.routeOptions.url (e.g., /api/users/:id)

This means /api/users/1, /api/users/2, and /api/users/999 all appear as a single /api/users/:id entry in the dashboard.

Aggregation

Raw request data is buffered in memory and flushed to the database every 5 seconds (configurable via collection.aggregationWindow). During each flush, LoadFlux computes:

• Request count and error count
• Latency percentiles: p50, p90, p95, p99
• Min/max/avg duration
• Status code distribution: 2xx, 3xx, 4xx, 5xx counts
• Total response bytes

The buffer uses a swap pattern — a new buffer is created before the old one is flushed, so there's zero blocking on the hot path.

Slow requests

Requests exceeding the slowRequestThreshold (default: 500ms) are highlighted in the Endpoints page under "Slow Requests". You can adjust this threshold in code or from the Settings page in the dashboard.

Error tracking

Responses with status codes >= 400 are logged in the error log with:

• Timestamp, method, and path
• Status code
• Error message (from response body if available)
• Stack trace (if available)
• Request duration

View these in the Errors page of the dashboard.

Dashboard search and filters

On the Endpoints page, a search box filters top endpoints, the endpoint metrics table, and slow requests (server-side search query parameter).

On the Errors page, search narrows the error log; you can also filter by HTTP status code for the current time range (status query parameter).

See REST API for all query parameters.

Excluding routes

To exclude specific routes from monitoring:

app.use(loadflux({
  excludeRoutes: ["/health", "/ready", "/metrics"],
}));

Excluded routes won't appear in the dashboard or affect any metrics.

Performance

The middleware adds < 0.5ms overhead per request:

• Express: Uses process.hrtime.bigint() for nanosecond timing, captures metrics in the res.finish event
• Fastify: Uses Fastify's built-in reply.elapsedTime and the onResponse hook

LoadFlux's own routes (/loadflux/*) are not counted in metrics.