When a bot stops responding there is an order to work through, from the control that risks nothing to the one that destroys the container. Work down it, not up.
1. Read the console first
Open the game panel from the service page and look at the console. A bot that has crashed almost always says why on its way out: an exception with a file and a line, a failed dependency, an invalid token, a rate limit. Restarting without reading this just means it crashes again in the same way. The console is also where you see whether the process is running at all.
2. Restart
The power controls are on the service page in the client area and in the game panel:
- Restart: stops the bot and starts it again. This is the right first move for a bot that is running but unresponsive, stuck on a reconnect, or has been sitting on a stale gateway session.
- Stop then Start: the same thing with a pause, useful when you want to change a file in between.
- Kill: stops the process immediately, without letting it shut down. Use it only when a stop does not work; anything the bot had not yet written to disk is lost.
3. Check the start command and the environment
If the bot will not start at all after an upload, it is usually one of these: the startup file points at a file that is not there, a dependency is missing from package.json or requirements.txt, or the token in the environment variables is wrong or has been regenerated in the Discord developer portal. All three are visible in the console output.
4. Reinstall, as a last step
Reinstall runs the installation script for the runtime again. It is the thing to try when the container's dependencies are in a state you cannot untangle by hand. Understand what it does first: files on the container may be changed or deleted, and it cannot be undone.
Before pressing it:
- Stop the bot.
- Take a backup and download it, so the copy is not inside the thing you are about to rebuild.
- Make sure your source exists somewhere else as well (a repository or your own machine).
Not the same thing: Change runtime
Changing the runtime between Node.js and Python rebuilds the container entirely and deletes everything on it. It is not a troubleshooting step; it is a decision about what language your bot is written in. Treat it as starting again from an empty container.
Still stuck?
Open a ticket from the service page and paste the console output as text. The ticket arrives with the service attached, and the output is almost always enough to say what is wrong.