API Health Monitor

A local API monitor built with JavaScript, Node.js, Express and SQLite. Checks HTTP status, response time and optional JSON values. Three consecutive failures open an incident; the next successful check resolves it.

Run

Requires Node.js 24 or newer.

npm ci
npm start

The API listens at http://localhost:3000. Data is saved to data/monitor.db. Set PORT or DB_PATH to override either default. Use npm run dev to restart on file changes.

Try it

# Create a monitor with a 60-second check interval
curl -X POST http://localhost:3000/monitors \
  -H 'Content-Type: application/json' \
  -d '{"name":"Example","url":"https://example.com","intervalSeconds":60}'

# List monitors and their current status
curl http://localhost:3000/monitors

# Run a check now and save the result
curl -X POST http://localhost:3000/monitors/ID/check

# View recent check results
curl http://localhost:3000/monitors/ID/checks

# View open and resolved incidents
curl http://localhost:3000/monitors/ID/incidents

Replace ID with the ID returned when creating a monitor.

Method	Route	Purpose
GET	/health	Service health
GET, POST	/monitors	List or create monitors
GET, PATCH, DELETE	/monitors/:id	Read, edit or delete a monitor
POST	/monitors/:id/check	Run a check now
GET	/monitors/:id/checks	Recent results
GET	/monitors/:id/incidents	Incident history

History routes accept ?limit=50 (maximum 100), newest first.

Optional fields: expectedStatus (200), timeoutMs (5000), maxResponseMs, enabled (true), and expectedJson, such as {"path":"data.ready","equals":true}. JSON comparisons preserve types. Intervals range from 10 seconds to 24 hours; timeouts from 100 to 30,000 ms and cannot exceed the interval.

Behaviour

• At most four checks run at once, with one per monitor. Busy manual requests return 409 or 429. Checks have no immediate retries.
• New monitors are checked on the next scheduler tick. Later checks wait one interval after completion. Pausing stops scheduled checks; manual checks still work.
• State and history survive restarts. Editing resets the failure streak and closes open incidents as configuration_changed. Results from older configurations are discarded.
• Up to 100 monitors, 1,000 results and 100 incidents per monitor are retained. Deleting a monitor removes its history.
• Targets must resolve only to public addresses. Connections use the validated address, redirects are rejected, and responses are capped at 64 KiB. Deadlines include DNS and response reading.

This is a single-process tool with no login or frontend. It binds to loopback and rejects remote browser origins. Checks run from the machine hosting it; monitoring stops when that process or computer stops.

Tests and demo

npm test
npm run demo

The demo runs a local fixture through the API, checker and database: success, wrong JSON, HTTP 503, timeout, recovery. It uses an isolated test transport; the normal server still blocks local targets. Tests cover validation, network failures, history, scheduling, incidents, persistence and shutdown. GitHub Actions runs both commands.