Telegram Bot Thread Example

Demonstrates an API server plus a Telegram bot's long-polling loop running on an independent, supervised worker_thread via runDetached().

Run it

export TELEGRAM_BOT_TOKEN=your_token_here
npm run dev -w examples-telegram-bot-thread

How it works

App server (app.ts)

A normal Nodalite API server with routes, middleware, etc.

Bot worker (bot-worker.ts)

A separate file that uses node:worker_threads to run a Telegram bot polling loop:

import { workerData, parentPort } from 'node:worker_threads';
import { Bot } from './bot.js';

const bot = new Bot(workerData.token);

async function run() {
  for await (const update of bot.poll()) {
    await bot.handle(update);
  }
}

run().catch((err) => {
  parentPort?.postMessage({ type: 'crash', error: err.message });
});

Detached runner

import { runDetached } from '@nodalite/workers';

runDetached('./bot-worker.js', {
  name: 'telegram-bot',
  workerData: { token: process.env.TELEGRAM_BOT_TOKEN },
  onCrash: (err) => console.error('Bot crashed, restarting:', err),
});

Key points

• A crash in the bot worker doesn't take down the HTTP server
• The worker auto-restarts with exponential backoff
• The bot and API share the same deployment unit — one process to monitor
• This pattern is for long-running servers only (not serverless)