Sentinel 🛰️

Self-hosted uptime monitoring that fits in a single binary.

I got tired of paying for status-page SaaS tools that do one thing: show a green dot. Sentinel checks your sites, APIs, databases, and raw TCP ports — all from a dark dashboard you own and run anywhere. No account, no subscription, no data leaving your network.

---

What it does

Paste any URL into the bar at the top. Hit Check once for a quick one-off result, or + Monitor to add it to the board permanently. Hover a card and click ✕ to stop watching it. That's the whole workflow.

Behind the scenes, every target gets checked concurrently on a configurable interval. Each card shows live latency history as a sparkline, a row of green/red tick bars for the last ~30 minutes, current uptime %, and TLS certificate days remaining. When something flips from up to down (or recovers), the incident lands in the event feed and fires a webhook alert — simultaneously, to as many channels as you've configured.

Supported check types:

URL format	How it's checked
https://mysite.com	HTTP GET — status code, latency, TLS cert expiry
192.168.1.10:3000 or localhost:5000	Same, with http:// assumed
tcp://host:22	Raw TCP dial — SSH, Redis, anything that listens on a port
mysql://user:pass@host:3306/db	Real MySQL connection + db.Ping()
postgres://user:pass@host/db	Real Postgres connection + db.Ping()

Alert channels (configured from the 🔔 panel in the dashboard — no YAML needed): Slack · Microsoft Teams · Google Chat · Discord · any custom webhook

---

Get started

Requires Go 1.22+. No database, no Docker, no other services.

git clone https://github.com/sagargoswami2001/sentinel.git
cd sentinel
go run ./cmd/sentinel serve

Open http://localhost:8080.

To build a binary you can copy to a server:

# Linux (from any machine)
GOOS=linux GOARCH=amd64 go build -o sentinel ./cmd/sentinel

# Windows
go build -o sentinel.exe ./cmd/sentinel

---

Running it

# Dashboard on :8080, checks every 30s
sentinel serve

# Tune the interval and port
sentinel serve --interval 10s --listen :9090

# One-shot check — prints a table, exits 1 if anything is down
# (useful in cron jobs or CI)
sentinel check

Press / on the dashboard to jump focus to the URL bar.

---

Configuration

Targets added via the browser are saved back to the config file automatically, so they survive restarts. You can also manage configs/sentinel.yaml by hand:

defaults:
  timeout_seconds: 10
  expect_status: 200

targets:
  - name: Production API
    url: https://api.mysite.com/health

  - name: Staging VM
    url: http://10.0.0.5:8000
    timeout_seconds: 3

  - name: Auth service (returns 401 when healthy)
    url: https://auth.internal.mysite.com
    expect_status: 401

  - name: SSH — build server
    url: tcp://build-host:22

  - name: Azure MySQL
    url: mysql://sentinel:password@mydb.mysql.database.azure.com:3306/app

Alerts

Open the 🔔 Notifications panel in the dashboard, pick a platform, paste the webhook URL, click Add. Send a test alert to confirm it works before you need it. Alerts fire on every up→down and down→up transition for every target on the board.

To keep webhook URLs out of committed YAML (recommended for public repos), use env vars instead:

SENTINEL_SLACK_WEBHOOK=https://hooks.slack.com/... sentinel serve

Where to get a webhook:

• Slack — api.slack.com/apps → Incoming Webhooks
• Teams — channel → ⋯ → Workflows → "Post to a channel when a webhook request is received"
• Google Chat — space settings → Apps & integrations → Webhooks
• Discord — channel → Edit → Integrations → Webhooks

Prometheus metrics

Sentinel exposes a /metrics endpoint on the same port:

curl http://localhost:8080/metrics

sentinel_target_up{name="Production API",url="..."}           1
sentinel_target_latency_seconds{name="Production API",url="..."} 0.043
sentinel_target_cert_days_left{name="Production API",url="..."}  56
sentinel_checks_total{name="Production API",url="...",status="up"} 24

Add to your prometheus.yml:

scrape_configs:
  - job_name: sentinel
    static_configs:
      - targets: ['localhost:8080']

JSON API

curl http://localhost:8080/api/status

{
  "last_run": "2026-08-06T14:30:00Z",
  "statuses": [
    {
      "name": "Production API",
      "url": "https://api.mysite.com/health",
      "up": true,
      "status_code": 200,
      "latency_ms": 43,
      "cert_days_left": 56
    }
  ]
}

---

Deploy

Render (free)

1. Push this repo to GitHub
2. render.com → sign in with GitHub → New → Web Service → select the repo
3. Render detects the Dockerfile automatically → pick Free → Deploy

Your dashboard is live at https://<app-name>.onrender.com in about two minutes. The free tier sleeps after 15 idle minutes — first visit takes ~30s to wake. For a permanent instance that's always on, Docker on any cheap VPS works well.

Docker

docker build -t sentinel .
docker run -d --name sentinel --restart unless-stopped \
  -p 8080:8080 \
  -v $(pwd)/configs/sentinel.yaml:/etc/sentinel/sentinel.yaml \
  sentinel

Or pull the image published automatically on every push to main:

docker run -d -p 8080:8080 \
  -v $(pwd)/configs/sentinel.yaml:/etc/sentinel/sentinel.yaml \
  ghcr.io/sagargoswami2001/sentinel:latest

The container reads the PORT environment variable, so it drops straight into Render, Railway, Fly.io, or any other PaaS without config changes.

---

Project layout

sentinel/
├── cmd/sentinel/        # thin CLI entrypoint — flag parsing, nothing else
├── internal/
│   ├── checker/         # HTTP, TCP, MySQL, Postgres health checks
│   ├── config/          # YAML load + save, default merging
│   ├── metrics/         # Prometheus gauge and counter definitions
│   ├── notify/          # Slack, Teams, GChat, Discord, webhook senders
│   ├── report/          # colored terminal table for sentinel check
│   └── server/          # HTTP server, templates, CSS — all go:embed'd
└── configs/sentinel.yaml

---

License

MIT — do whatever you want with it.