This is your uploaded project (server.js, monitor.sh, index.html, Dockerfile) with three functional bugs fixed and a couple of production hardening tweaks. Nothing about your architecture changed.

What was wrong and what I changed

index.html was fetching the wrong URL. const API_URL = ""; then fetch(API_URL) calls the current page (/), not /health — so it was trying to JSON.parse() an HTML page and always fell into the catch block ("Offline"), even when the API was healthy. Fixed to fetch('/health').
server.js had no route serving index.html. The dashboard needs to be reachable somewhere. I added express.static('public') and moved index.html into public/, so the dashboard is served at / and the plain-text welcome message moved to /api so the two don't collide.
monitor.sh referenced $response_time, which was never set — the response-time column was always empty. Fixed by measuring it directly with curl -w "%{time_total}". Also added a fallback JSON parser for when jq isn't installed, and an optional CSV log file.
Dockerfile ran npm init -y && npm install on every build, with no lockfile/manifest — slow, non-reproducible, and re-downloads dependencies on every rebuild even if nothing changed. Added a real package.json and split COPY package.json from COPY server.js so Docker caches the npm install layer. Also switched node:20-alpine3.16 (an older, unsupported Alpine point release) to node:20-alpine (current supported tag).

Files

.
├── Dockerfile
├── package.json
├── server.js
├── public/
│   └── index.html      ← dashboard, served at /
└── monitor.sh           ← external checker, run it against the deployed URL

Run locally

npm install
npm start
# open http://localhost:3000        -> dashboard
# open http://localhost:3000/health -> raw JSON

In another terminal, point the monitor at it:

chmod +x monitor.sh
./monitor.sh http://localhost:3000 30

Build & run with Docker

docker build -t api-health-demo .
docker run -d -p 3000:3000 --name api-health-demo api-health-demo

`/health` response shape

{
  "status": "Online",
  "uptime_seconds": 42,
  "request_count": 17,
  "response_time_ms": 3,
  "timestamp": "2026-07-27T20:10:00.000Z"
}

Deploy to ghaymah.systems

See DEPLOY_GHAYMAH.md for the step-by-step (build → push → create service → set port 3000 → point monitor.sh at the live URL).

README.md