Deploy a Personal chatgpt-telegram-bot

ChatGPT usage can be unstable. To avoid downtime impacting work, I set up a private Telegram bot as a backup. After some research, I chose to deploy the excellent chatgpt-telegram-bot.

Setup steps

1. In Telegram, talk to @BotFather to create a bot.

2. In Telegram, talk to @useridinfobot to get your user ID (not your display name).

3. Choose a HK or overseas server and pull the image.

   docker pull n3d1117/chatgpt-telegram-bot:latest
   

4. Create a .env file and set environment variables. Reference: https://github.com/n3d1117/chatgpt-telegram-bot/blob/main/.env.example

   The four main settings:

   # Your OpenAI API key
   OPENAI_API_KEY=XXX
   
   # Your Telegram bot token obtained using @BotFather
   TELEGRAM_BOT_TOKEN=XXX
   
   # Telegram user ID of admins, or - to assign no admin
   ADMIN_USER_IDS=ADMIN_1_USER_ID,ADMIN_2_USER_ID
   
   # Comma separated list of telegram user IDs, or * to allow all
   ALLOWED_TELEGRAM_USER_IDS=USER_ID_1,USER_ID_2
   

5. Start the container

   docker run  -d --name "chatgpt-telegram-bot" --env-file ./.env n3d1117/chatgpt-telegram-bot:latest
   

6. Now you can talk to your bot in Telegram.

Auto‑restart on config changes

With a plain docker run container, updating the .env file won’t auto‑reload the container. To recreate it on changes, I used inotify-tools (CentOS 7 example below):

1. sudo yum -y install inotify-tools
2. Create watch_env.sh

#!/bin/sh
while true; do
  inotifywait -e modify .env
  echo ".env file has been modified. Restarting the container..."
  docker stop chatgpt-telegram-bot
  docker rm chatgpt-telegram-bot
  docker run -d --name "chatgpt-telegram-bot" --env-file /root/chatgpt-telegram-bot/.env n3d1117/chatgpt-telegram-bot:latest
done

2. Create watch-env.service

sudo vi /etc/systemd/system/watch-env.service

Service configuration:

[Unit]
Description=Watch and restart Docker container on .env file changes

[Service]
Type=simple
ExecStart=/bin/bash /root/chatgpt-telegram-bot/watch_env.sh
Restart=always
User=root

[Install]
WantedBy=multi-user.target

3. Enable and start the service

sudo systemctl daemon-reload

sudo systemctl enable watch-env.service

sudo systemctl start watch-env.service

GPT‑4

After getting access approval, set OPENAI_MODEL=gpt-4 in .env.

Apply here: https://openai.com/waitlist/gpt-4-api

Final Thoughts

That’s it — a personal bot is up. It’s paid, but relatively stable. Hoping to get API access for GPT‑4 soon to make it even better.