Hello Railway Support Team,

I am experiencing an issue with long polling in my Telegram bot running on your platform.

Symptoms:

- The bot crashes with the error: "HTTP Client says - Request timeout error".

- Telegram API works fine from my local machine using curl:

curl -v https://api.telegram.org/bot<TOKEN>/getMe

Response: 200 OK, JSON with bot information.

- The timeouts occur only on Railway, indicating a possible network or infrastructure problem.

Suspicions:

- Potential issues with long polling / persistent TCP connections.

- Possible IPv6 / routing problems.

- Free tier containers may be unstable with long-lived HTTP connections.

Questions:

1. Are there any known issues with long polling or TCP connections on your infrastructure?

2. Could you check if there is packet loss or restrictions for my project?

Thank you for your help.

Best regards,

Hello!
I have the same issue.
Failed to contact Telegram while validating the bot token

All my work bots suddenly stopped working.

Railway, please HELP

+1

Same here

Same

Having the same issue.

Deployed into us-west-2, using Grammy for NodeJS

Same here

Guys, please click on the arrow button on the initial post, hope the issue will be prioritised.
Thanks!

Same here

.

bora railway, agita ai

Teu bot é de que? Kkk

Here is a similar issue: https://station.railway.com/questions/python-telegram-bots-timing-out-on-railw-c424f0c7

Essa carniça aqui: @SuperListaBot , divulgação de grupos

O meu é o B

Same here

Pow legal em , deve ter bastante público

por que pessoal gosta de criar um fake kkkk

is anyone going to manage this issue????

we depend on your beloved railway

2 hours downtime for services with active marketing promotion... :(

see support next year

It's not ideal, but I just found a workaround for this in the meantime.

Create a Cloudflare account, create a new Worker and set the code to the below to make a very cheap/easy proxy using Cloudflare which has a very generous free tier.

export default {
  async fetch(request) {
    const url = new URL(request.url);
    
    // Rewrite the URL to point to Telegram API
    const telegramUrl = `https://api.telegram.org${url.pathname}${url.search}`;
    
    // Forward the request
    return fetch(telegramUrl, {
      method: request.method,
      headers: request.headers,
      body: request.method !== 'GET' ? request.body : undefined,
    });
  },
};

Then in your code, whatever framework you're using, set the URL of the new worker as your proxy/callback URL.

For instance, using Grammy for NodeJS the following resolves by using Cloudflare as a simple proxy:

const bot = new Bot(process.env.TELEGRAM_BOT_TOKEN, {
  apiRoot: process.env.TELEGRAM_API_PROXY_URL,
});

## Confirmed Issue - Railway-Wide Telegram API Connectivity Problem

English:

I can confirm this is affecting multiple users across Railway. I'm experiencing the identical issue with n8n workflows running Telegram Bot API nodes.

Key Findings:

- DNS resolution works correctly (149.154.167.220)

- Other HTTP/HTTPS requests from Railway function normally

- Only connections to api.telegram.org:443 timeout

- Timeout occurs at TCP connection phase, NOT TLS/HTTP layer

- Affects ALL Telegram Bot API endpoints

This appears to be a Railway network/firewall issue blocking outbound HTTPS connections specifically to Telegram's IP ranges (149.154.167.0/24 and related blocks). The same bots work perfectly on local machines and other hosting providers.

Temporary workaround: Anonymous user above shared a Cloudflare Workers proxy solution - excellent idea for urgent production needs!

To everyone affected: Stay strong, hang in there! We're all in this together. Hopefully Railway team will resolve this infrastructure issue soon.

---

Русский:

Подтверждаю - проблема массовая, затрагивает множество пользователей Railway. У меня идентичная ситуация с n8n workflows и Telegram Bot API.

Основные выводы:

- DNS резолв работает корректно (149.154.167.220)

- Другие HTTP/HTTPS запросы с Railway работают нормально

- Только подключения к api.telegram.org:443 таймаутятся

- Таймаут на уровне TCP connect, НЕ TLS/HTTP

- Затронуты ВСЕ эндпоинты Telegram Bot API

Это похоже на проблему сети/файрвола Railway, блокирующую исходящие HTTPS-соединения именно к IP-диапазонам Telegram (149.154.167.0/24 и связанные блоки). Те же боты отлично работают на локальных машинах и других хостингах.

Временное решение: Anonymous выше поделился решением через Cloudflare Workers proxy - отличная идея для срочных production задач!

Всем пострадавшим: Держимся, братцы! Мы все в одной лодке. Надеюсь, команда Railway скоро решит эту инфраструктурную проблему.

Удачи всем! Good luck everyone!

Держимся, братцы! ахаха :))) надеюсь Railway это переведет :))

запрацювало!

смысле само? или ты что то сделал?

само

у меня пока нет

Сделал редеплой, все та же хрень - не работает

О! заработало!

Где то я уже это слышал)

Кстати у всех с октября за N8N дороже получается платить?) (оффтоп)

what did you do

I did nothing. Bot became alive suddenly in chat and strated to answer missed questions for 2 hours

Some Telegram bots have returned, others have not.

правда причем не после нового деплоя а просто )

My bots have returned to work

One of them I had to redeploy

и перестало опять :((( And it stopped again

мої три працюють...

three of mine still work

post in english please

сrashed again

Experiencing same issue. Please resolve it. It does not appear to be a Telegram API issue, although others are experiencing similar delays. One request was able to eke through somehow.

Crashed again

e lá vamos nós de novo kkkk

Aqui também!! Vou cancelar minha assinatura tem dias que está assim

E vai pra onde? Saí da squarecloud pq bugava e aqui estou

aqui é bem melhor do que a square

Também acho

A equipe do railway ainda não deu nenhum parecer sobre esse problema?!

não

Anyone able to get the connection to the TG API to work? It works sometimes and not stable

Faça isso que tudo se resolverá:

Não é o ideal, mas encontrei uma solução alternativa provisória.

Crie uma conta na Cloudflare, crie um novo Worker e defina o código abaixo para criar um proxy muito barato e fácil usando a Cloudflare, que possui um plano gratuito bastante generoso.

export default {

async fetch(request) {

const url = new URL(request.url);

// Rewrite the URL to point to Telegram API

const telegramUrl = `https://api.telegram.org${url.pathname}${url.search}`;

// Forward the request

return fetch(telegramUrl, {

method: request.method,

headers: request.headers,

body: request.method !== 'GET' ? request.body : undefined,

});

},

};

Em seguida, no seu código, independentemente do framework que estiver usando, defina a URL do novo worker como a URL do seu proxy/callback.

Por exemplo, usando o Grammy para NodeJS, o seguinte código é resolvido usando o Cloudflare como um proxy simples:

const bot = new Bot(process.env.TELEGRAM_BOT_TOKEN, {

apiRoot: process.env.TELEGRAM_API_PROXY_URL,

});

https://station.railway.com/questions/bot-error-ef3e0945#nrsa

Hello all,

Sorry for the delayed ack here on our side. It seems like someone was sending some outbound traffic causing our IP set to be ratelimited. We don't have a relationship with the TG team although we should, we're going to see what we can do in the meantime. In the shortterm, proxying your API calls would be the best workaround for now.

Do you have any ETA when everything will be fine?

Same Issue Here.

AGAIN!

Contact TG support, the problem is massive, we really look forward to a solution https://telegram.org/support

@Railway team,

Any updates? :)

Hi folks, we are unfortunately still waiting to hear back from Telegram (resolution is largely outside our control). As Angelo mentioned, the most reliable workaround in the meantime would be to use a proxy.

We know this is frustrating, especially if you're running production services. We'll post an update here as soon as we have more information.

A note to the community and folks on this thread: if anyone here has been able to find a workaround using a proxy, could you please share a detailed technical explanation of how you did it?

If it works for others as a holdover while Telegram gets back to us, we will be happy to compensate your preferred workspace $50 in Railway creditsfor helping out your fellow users

Best, The Railway Team

Essa é a solução temporária:

Não é o ideal, mas encontrei uma solução alternativa provisória.

Crie uma conta na Cloudflare, crie um novo Worker e defina o código abaixo para criar um proxy muito barato e fácil usando a Cloudflare, que possui um plano gratuito bastante generoso.

exportar padrão {

async fetch(request) {

const url = new URL(request.url);

// Reescreva a URL para apontar para a API do Telegram

const telegramUrl = `https://api.telegram.org${url.pathname}${url.search}`;

// Encaminhar a solicitação

retornar fetch(telegramUrl, {

método: método de solicitação,

cabeçalhos: request.headers,

corpo: método da requisição !== 'GET' ? corpo da requisição : indefinido,

});

},

};

Em seguida, no seu código, independentemente do framework que você estiver usando, defina a URL do novo trabalhador como a URL do seu proxy/callback.

Por exemplo, usando o Grammy para NodeJS, o seguinte código é resolvido usando o Cloudflare como um proxy simples:

const bot = new Bot(process.env.TELEGRAM_BOT_TOKEN, {

apiRoot: process.env.TELEGRAM_API_PROXY_URL,

});

Hello!

We're acknowledging your issue and attaching a ticket to this thread.

We don't have an ETA for it, but, our engineering team will take a look and you will be updated as we update the ticket.

Please reply to this thread if you have any questions!

🛠️ The ticket Outbound connection issues in EU region has been marked as triage.

Hi Chandrika,

I posted previously in this thread: https://station.railway.com/questions/bot-error-ef3e0945#nrsa

About using Cloudflare workers as a proxy. I've been using this since yesterday and it has been working consistently.

Amazing, thank you so much peppay! I'll go ahead and share this with a few other users not on this thread in the meantime while we're waiting to hear back from Telegram

($50 have been added as credits to your workspace)

É um absurdo o que a gente paga por uso nesses planos e ainda tem esses problemas!! Lamentavél a ferrovia está enferrujada!

I think affected customers should get some compensation as well, i have a production app, with some angry clients not getting notifications, my entire app is telegram based.

In my view, I support the idea that compensation should be provided to all affected parties. Today is New Year’s Day, and many people had their most important orders disrupted. I reported the problem yesterday, but it only started being addressed 12 hours later. The instructions shared by the user are incomplete, and it is currently quite difficult to understand and work with them. If everything had been explained step by step, it would have at least partially resolved the issue.

Também acho fiquei um ou dois dias com esse problema perdi muitas vendas e esse mês a cobrança vem alta ainda por uso.. sem nem mesmo usar direito!!

FULL WORK GUIDE

## Step 1: Install Wrangler CLI

```bash

npm install -g wrangler

```

## Step 2: Create Worker Project

```bash

mkdir telegram-api-proxy && cd telegram-api-proxy

mkdir src

```

## Step 3: Create Files

*src/index.js:**

```javascript

export default {

async fetch(request) {

const url = new URL(request.url);

const telegramUrl = https://api.telegram.org${url.pathname}${url.search};

const headers = new Headers(request.headers);

headers.delete('Host');

return fetch(telegramUrl, {

method: request.method,

headers: headers,

body: request.method !== 'GET' && request.method !== 'HEAD' ? request.body : undefined,

});

},

};

```

*wrangler.toml:**

```toml

name = "telegram-api-proxy"

main = "src/index.js"

compatibility_date = "2025-01-01"

```

## Step 4: Deploy

```bash

wrangler login

wrangler deploy

```

Copy the Worker URL (e.g., https://telegram-api-proxy.your-subdomain.workers.dev)

## Step 5: Update Your Bot Code

Your bot code should read TELEGRAM_API_PROXY_URL from environment variables and use it if set.

### Python (python-telegram-bot)

```python

from telegram.request import HTTPXRequest

from telegram.ext import Application

from app.core.config import settings # Your settings module

builder = Application.builder().token(settings.TELEGRAM_BOT_TOKEN)

# Use Cloudflare Worker proxy if configured

if settings.TELEGRAM_API_PROXY_URL:

from telegram.request import HTTPXRequest

proxy_url = settings.TELEGRAM_API_PROXY_URL.rstrip('/')

custom_request = HTTPXRequest(

connection_pool_size=8,

connect_timeout=30.0,

read_timeout=30.0,

write_timeout=30.0,

pool_timeout=10.0,

base_url=proxy_url # Worker will forward to api.telegram.org

)

builder = builder.request(custom_request)

logger.info(f" Using Cloudflare Worker proxy: {proxy_url}")

application = builder.build()

```

**In your config file (e.g., config.py):**

```python

from pydantic_settings import BaseSettings

from typing import Optional

class Settings(BaseSettings):

TELEGRAM_BOT_TOKEN: str

TELEGRAM_API_PROXY_URL: Optional[str] = None # Cloudflare Worker URL (from Railway Variables)

```

**Note:** The bot reads TELEGRAM_API_PROXY_URL from Railway Variables. If the variable is set, it uses the proxy. If not set, it uses direct connection to Telegram API.

## Step 6: Set Environment Variable in Railway

1. Go to Railway Dashboard

2. Open your project

3. Navigate to **Variables** (or **Environment Variables**)

4. Click **"+ New Variable"**

5. Add:

- **Name:** TELEGRAM_API_PROXY_URL

- **Value:** https://telegram-api-proxy.your-subdomain.workers.dev (your Worker URL from Step 4)

6. Click **Save**

Railway will automatically restart your service after adding the variable.

## Step 7: Verify It's Working

Railway automatically restarts your service after adding the variable. Check Railway logs for:

```

[CREATE_BOT] Используется Cloudflare Worker прокси: https://telegram-api-proxy.your-subdomain.workers.dev

```

or in English:

```

Using Cloudflare Worker proxy: https://telegram-api-proxy.your-subdomain.workers.dev

```

If you see this message, the proxy is working! Test your bot by sending a command.

## Troubleshooting

**Bot not responding?**

1. Check Railway logs for errors

2. Verify TELEGRAM_API_PROXY_URL is set correctly in Railway Variables

3. Test Worker URL directly:

```bash

curl https://telegram-api-proxy.your-subdomain.workers.dev/bot<your-token>/getMe

```

4. Check Worker logs in Cloudflare Dashboard → Workers & Pages → Your Worker → Logs

**Need to disable proxy?**

1. Go to Railway Dashboard → Your Project → Variables

2. Delete the TELEGRAM_API_PROXY_URL variable

3. Railway will restart the service automatically

4. Bot will use direct connection to api.telegram.org

You can use the workaround steps from alpacco (THANK YOU BTW!) or- I spent some time trying to get this experience as one click as possible.

Update: the Railway team is still trying to get in contact with the TG team. We now have a one-click deploy that uses CF Workers to set up a proxy to allow you to proxy requests to get around the connection issue. You can use the above steps from

https://railway.com/deploy/tg-cloudflare-worker-proxy-workaround?referralCode=N8f7Nb&utm_medium=integration&utm_source=template&utm_campaign=generic

I agree.

Any type of proxy is a workaround, I wasn’t ready to do on New Years eve.

I have my services here and expect that all of them will work.

I had active promotion for 2 my bots and the whole traffic was lost due to the Railway’s infrastructure problem, so I expect that Railway team would compensate all the problems, I had.

This has been going for too long, it's caused so many issues! Any updates?

We are waiting to hear back from Telegram.