Description of the Issue:

I'm deploying a Telegram bot using Node.js and the Telegraf library on Railway. The bot works fine locally and also when deployed to Render, but fails on Railway when trying to set a webhook.

The failure seems to be caused by a semicolon (;) being appended automatically to the webhook URL constructed using process.env.APP_URL, even though the environment variable in Railway does not contain a semicolon in the value.

Error Message and Description:

When setting the webhook via:

CopyEdit

await bot.telegram.setWebhook(`${process.env.APP_URL}/webhook`);

I get this error:

CopyEdit

TelegramError: 404: Not Found payload: { url: 'https://bot-test-production-1293.up.railway.app/webhook'; }

Note the semicolon at the end of the webhook URL:

'https://bot-test-production-1293.up.railway.app/webhook';

This is not present in my variable value. It breaks the Telegram API call.

I also tested this via manual curl:

curl -X POST https://api.telegram.org/bot<my_token>/setWebhook \ -H "Content-Type: application/json" \ -d '{"url": "https://bot-test-production-1293.up.railway.app/webhook"} Telegram returned {"ok":true}, proving the endpoint is correct and reachable when the semicolon is not present.

Build/Deploy Logs:

There’s no error during build. Error occurs during runtime when the webhook is being set:

Failed to launch bot: TelegramError: 404: Not Found payload: { url: 'https://bot-test-production-1293.up.railway.app/webhook'; }

This shows the invalid semicolon.

Code Repository (If applicable):

[Private repo] – But I can provide a stripped-down minimal version if needed. The important code snippet:

const WEBHOOK_PATH = "/webhook"; const WEBHOOK_URL = $`{process.env.APP_URL}${WEBHOOK_PATH}`; await bot.telegram.setWebhook(WEBHOOK_URL);

Conclusion and Request:

I believe Railway is injecting a semicolon when interpolating or exporting APP_URL, even though it’s not in the variable editor. This causes third-party API integrations (like Telegram webhooks) to break silently.

Please help confirm:

Is this a known issue?

Can it be fixed or worked around?

Is there an internal sanitization step I need to disable?

This issue only happens on Railway and cost me a lot of debugging time. Happy to assist with more info or test cases.