Notifications Uptime March 13, 2026 · 6 min read

Get Telegram Bot Downtime Notifications Before Your Users Do

The worst way to find out your bot is down is a message from a user. Here are three approaches to get notified of downtime instantly — from a 10-minute DIY script to a fully automated repair system that fixes it before you even know it happened.

Why default OpenClaw monitoring isn't enough

OpenClaw logs events, but it doesn't alert you when the gateway goes down. The gateway process doesn't know to send you a message — because it's the thing that sends messages, and it's dead.

You need something external to the gateway watching it. Here are three options in order of complexity:

Method 1: Self-pinging cron alert

🔧 Cron + Telegram Bot API Easy — 10 minutes

A cron job runs every few minutes, checks if the gateway is alive, and sends you a Telegram message if it's not. You need a second bot token to send the alert (can't use the down bot to tell you it's down).

alert.sh

#!/bin/bash
# Get these from @BotFather (alert bot, separate from your main bot)
ALERT_TOKEN="123456:your-alert-bot-token"
ALERT_CHAT="987654321"  # Your personal Telegram chat ID (get from @userinfobot)

if ! pgrep -f "openclaw-gateway" > /dev/null 2>&1; then
  MSG="⚠️ *OpenClaw bot is down* — $(hostname) — $(date '+%H:%M UTC')"
  curl -s -X POST "https://api.telegram.org/bot${ALERT_TOKEN}/sendMessage" \
    -d "chat_id=${ALERT_CHAT}" \
    -d "text=${MSG}" \
    -d "parse_mode=Markdown" > /dev/null
fi

Add to crontab (runs every 5 minutes)

*/5 * * * * /usr/local/bin/openclaw-alert.sh

Pros

Free
No dependencies
10 minutes to set up

Cons

5 minute detection lag
Alert only — no repair
You still need to fix it

Method 2: Health endpoint + external monitor

📡 Uptime robot / Better Uptime Medium — 30 minutes

If your OpenClaw bot exposes an HTTP health endpoint, external services like UptimeRobot (free) or Better Uptime ($7/mo) can monitor it every minute and alert you via Telegram, email, or SMS.

Add a health endpoint to your OpenClaw workspace

Add this to your OpenClaw config or a companion Express server:

health-server.js

// Companion health server for OpenClaw
const http = require('http');
const { execSync } = require('child_process');

const server = http.createServer((req, res) => {
  if (req.url === '/health') {
    try {
      // Check if gateway process is running
      const pid = execSync('pgrep -f openclaw-gateway').toString().trim();
      res.writeHead(200, {'Content-Type': 'application/json'});
      res.end(JSON.stringify({ ok: true, pid }));
    } catch {
      res.writeHead(503, {'Content-Type': 'application/json'});
      res.end(JSON.stringify({ ok: false, error: 'gateway not running' }));
    }
  } else {
    res.writeHead(404); res.end();
  }
});

server.listen(8080, '127.0.0.1', () => {
  console.log('Health server on http://localhost:8080/health');
});

Then monitor https://your-server.com/health in UptimeRobot. When it returns non-200, you get alerted.

Pros

1-minute detection
Free tier available
Email + SMS + Telegram

Cons

Still manual repair
Extra server to maintain
False positives if server slow

Method 3: Mechanic Bot — alert + auto-repair

🔧 Mechanic Bot Full automation

The above approaches tell you the bot is down. You still have to fix it. Mechanic Bot does both: detects within 60 seconds, attempts auto-repair, and then notifies you — with the outcome already included.

What the notification looks like:

Telegram notification from Mechanic
✅ Fixed in 47s — MyBot is back online

Your bot went offline at 02:14 UTC.
Mechanic detected and repaired it automatically.

What happened: Config file had a syntax error
What was done: Backup → deep-merge repair → restart
Method: config_check_restart (attempt 1 of 5)

↗ View repair details

You get this message instead of the alert that woke you up. The repair already happened.

Setup (4 minutes)

Linux/VPS

curl -fsSL https://mechanicbot.io/onboard.sh | sudo bash -s -- \
  --token YOUR_INSTALL_TOKEN \
  --email you@example.com

After install, link your Telegram in the dashboard to receive repair notifications in your personal Telegram chat.

Pros

Fixes it before you reply
60-second detection
Handles config corruption
Audit trail of every repair

Cons

$9/mo (vs. free cron)
Requires install on server

Which approach should you use?

It depends on how much the downtime costs you:

Hobby bot, few users: Method 1 (free cron) is fine. 5-minute lag is acceptable.
Bot with paying/active users: Method 2 gives you 1-minute detection. Still manual repair, but you know fast.
Bot running for clients / business: Method 3. Downtime costs more than $9/mo in reputation + time to fix.

The rule of thumb

If you'd be embarrassed by 2 hours of undetected downtime, you need automated repair — not just alerts.

Get alerted when it's already fixed.

Mechanic Bot monitors your OpenClaw bot 24/7. When it goes down, Mechanic fixes it and sends you a notification with the outcome — not a wake-up call.

Start 14-day free trial → Bot down right now?