Telegram Downtime(r)

✦ Imagine this scenario, you've deployed on a Friday evening (rookie mistake), went to bed and woke up realizing your beloved Telegram bot, due to some unhandled error, has been down for the past 9 hours with no sort of announcement sent out to the users. Very, very awkward.

Note

⌘ This was initially built for personal projects deployed on Railway and may not work as expected on other hosting providers, you will need to tweak the code to fit this project to your needs. Pull-requests to make this bot more extensible are always welcome.

This is a very small server which will listen for incoming downtime updates (essentially HTTP requests) which are sent by your main bot whenever it comes across, for example, an uncaught exception (via process.on('uncaughtException')). Once a request is received, the downtime will be started or stopped according to the body sent with the request.

Here's a simple method of sending downtime updates to this server:

/**
 * Fired whenever the bot unexpectedly goes down (e.x. via an `uncaughtException` event)
 * or instead, comes back up. This function will send a status update to the
 * downtime handler which will (if the bot is experiencing downtime) take over
 * and respond to incoming messages with a downtime notice.
 *
 * You can also fire this function manually to represent maintenance or similar.
 * Make sure you have a way of stopping the downtime, perhaps by restarting the
 * bot. Sending manual requests is NOT possible as they are sent to an internal
 * network.
 *
 * @async
 * @function sendDowntimeUpdate
 * @param down - Whether the bot is currently down or not.
 * @returns Nothing.
 *
 * @example
 * ```ts
 * process.on('uncaughtException', async (err) => {
 *     console.log("I told you to not deploy on a Friday!", err);
 *     await sendDowntimeUpdate(true); // tell the downtime server that the bot is down.
 * });
 * ```
 */
export async function sendDowntimeUpdate(down: boolean): Promise<void> {
    /**
     * The **internal** domain, located on an **internal** network which would restrict any
     * manual request(s) made to this endpoint. Private networks are secure out of the box.
     */
    const networkDomain = `http://${config.DOWNTIME_HOST}:${config.DOWNTIME_PORT}`;
    /** The endpoint to which downtime updates are sent. */
    const downtimeEndpoint = new URL('downtime', networkDomain).href;
    /**
     * Data to pass alongside with the request. This is only the request's body, not the
     * configuration. Headers, cookies, etc. must be set within the request itself.
     */
    const updateBodyData = { down }; // new downtime state.

    // send the request to the downtime handler.
    await axios.post(downtimeEndpoint, updateBodyData).catch((err) => {
        // handle an error accordingly.
        console.error(
            "Couldn't send the downtime update, make sure your downtime handler is reachable.",
            err
        );
    });
}

Using whatever method of error handling, whether the native process.on() or a custom implementation, simply call this function and pass a boolean to send a downtime update. This bot will handle everything else.

Environment variables:

Here's a breakdown of every environment variable used by the project.

Variable	Short Explanation.	Recommended Value
BOT_TOKEN	The same token you use for the MAIN BOT!	Your Token
DATABASE_URL	The URL of the MongoDB database where your users are stored.	Your Database
DB_NAME	The name of the database.
DB_COLLECTION	The name of the collection within the database.
ONCALL_ADMIN	The Telegram ID of the user who should be notified when your bot goes down.	Yourself
PROJECT_ID	Internal.	Railway Project
SERVER_PORT	The port to bind the downtime server to.	443
SUPPORT_CHAT	A short link through which users can contact you for support during downtime.
FORCE_DOWNTIME	Self-explanatory, will create a downtime without any requests.	false
DOWNTIME_DELAY	How many seconds to wait before starting the downtime (incase your main bot can self-recover).	60
BLOCK_FIELD	The field within the database which controls the block status for a user.

Deploy on Railway

As I've mentioned previously, this project is supposed to be deployed on the Railway platform.

You can use this template to quickly deploy this downtime handler, literally in a few clicks. You'll make sure your bot no longer suffers from downtime and I'll receive a small referral cut, sounds great, right? Make sure you've got a Railway account setup.

As for chat member updates:

✦ Don't forget that people will still block your bot, everytime that happens, this bot will update their block status in the connected database. You will have to update the field which represents a user's block status.