Here are some instructions to get esmBot up and running from source.
??? check "Recommended system requirements"
- 64-bit CPU/operating system
- Quad-core CPU or better
- 1GB or more of RAM
- Linux-based operating system or virtual machine ([Ubuntu 22.04 LTS](https://ubuntu.com/download/server) or [Fedora 36](https://getfedora.org/) are recommended)
!!! warning
If you want to run the bot on Windows, [Windows Subsystem for Linux](https://docs.microsoft.com/en-us/windows/wsl/install-win10) is recommended. This guide is somewhat Linux-centric, so for now you're mostly on your own if you decide not to use WSL.
If you have any further questions regarding setup, feel free to ask in the #support channel on the [esmBot Support server](https://esmbot.net/support).
!!! tip
You can run the bot using Docker for a somewhat simpler setup experience. [Click here to go to the Docker setup guide.](https://docs.esmbot.net/docker)
### 1. Install the required native dependencies.
Choose the distro you're using below for insallation instructions.
=== "Debian/Ubuntu"
These instructions apply to Debian version 12 (bookworm) or Ubuntu version 22.04 (jammy) or later.
On older Debian/Ubuntu versions, you may need to install some of these packages (notably libcgif-dev and meson) through alternative methods.
=== "Fedora/RHEL"
These instructions apply to Fedora 36/RHEL 9 or later.
Some of these packages require that you add the RPM Fusion and/or EPEL repositories. You can find instructions on how to add them [here](https://rpmfusion.org/Configuration).
You'll also need to install [`ttf-ms-win10-auto`](https://aur.archlinux.org/packages/ttf-ms-win10-auto/) from the AUR.
***
### 2. Install libvips.
[libvips](https://github.com/libvips/libvips) is the core of esmBot's image processing commands. Version 8.13.0 or higher is recommended because it contains fixes to GIF handling and support for the freeze command; however, this version isn't packaged for most distros yet. To fix this, you'll need to build libvips from source.
!!! note
Alpine, Arch, RHEL **(not Fedora!)**, and Ubuntu 22.10 (Kinetic Kudu) users can skip this step, since these distros now have 8.13.0 packaged.
If that command finishes with no errors, you can compile and install it:
```sh
cd build
meson compile
sudo meson install
```
***
### 3. Install Node.js.
Node.js is the runtime that esmBot is built on top of. The bot requires version 16 or above to run.
First things first, we'll need to install pnpm, the package manager used by the bot. Run the following to install it:
```sh
curl -fsSL https://get.pnpm.io/install.sh | sh -
```
Once you've done that, continue with the instructions for your operating system below.
=== "Debian/Ubuntu"
You'll need a more recent version than what's provided in most Debian/Ubuntu-based distros. You can add a repository that contains a supported version by running this command:
After that, you can install Node.js with this command:
```sh
sudo apt-get install nodejs
```
=== "Fedora/RHEL"
```sh
sudo dnf install nodejs
```
=== "Alpine"
```sh
doas apk add nodejs
```
=== "Arch/Manjaro"
```sh
sudo pacman -S nodejs
```
***
### 4. Set up the database.
esmBot officially supports two database systems: SQLite and PostgreSQL. While SQLite is smaller and requires no initial setup, PostgreSQL has better performance (especially in large environments).
!!! tip
If you're new to databases and self-hosting, choose SQLite.
If you would like to use the SQLite database, no configuration is needed and you can move on to the next step.
If you would like to use the PostgreSQL database, view the setup instructions [here](https://docs.esmbot.net/postgresql) and come back here when you're finished.
***
### 5. Clone the repo and install the required Node modules.
Lavalink is the audio server used by esmBot for soundboard commands and music playback. If you do not plan on using these features, you can safely skip this step.
!!! warning
There are websites out there providing lists of public Lavalink instances that can be used with the bot. However, these are not recommended due to performance/security concerns and missing features, and it is highly recommended to set one up yourself instead using the steps below.
Lavalink requires a Java (11 or later) installation. You can use [SDKMAN](https://sdkman.io) to install Eclipse Temurin, a popular Java distribution:
You'll need to run Lavalink alongside the bot in order to use it. There are a few methods to do this, such as the `screen` command, creating a new systemd service, or simply just opening a new terminal session alongside your current one.
***
### 7. Configure the bot.
Configuration is done via environment variables which can be specified through a `.env` file. Copy `.env.example` to get a starter config file:
```sh
cp .env.example .env
```
!!! tip
If you can't see either of these files, don't worry - Linux treats files whose names start with a . as hidden files.
To edit this file in the terminal, run this command:
```sh
nano .env
```
This will launch a text editor with the file ready to go. Create a Discord application [here](https://discord.com/developers/applications) and select the Bot tab on the left, then create a bot user. Once you've done this, copy the token it gives you and put it in the `TOKEN` variable.
When you're finished editing the file, press Ctrl + X, then Y and Enter.
An overview of each of the variables in the `.env` file can be found [here](https://docs.esmbot.net/config).
***
### 8. Run the bot.
Once everything else is set up, you can start the bot like so:
```sh
pnpm start
```
If the bot starts successfully, you're done! You can invite the bot to your server by generating an invite link under OAuth -> URL Generator in the Discord application dashboard.
!!! note
You will need to select the `bot` and `applications.commands` scopes.
The following permissions are needed in most cases for the bot to work properly:
PostgreSQL isn't running, you should be able to start it with `sudo systemctl start postgresql`. If you don't intend to use PostgreSQL, you should take another look at your `DB` variable in the .env file.
??? faq "Gifs from Tenor result in a "no decode delegate for this image format" or "improper image header" error"
Tenor GIFs are actually stored as MP4s, which libvips can't decode most of the time. You'll need to get a Tenor API key from [here](https://developers.google.com/tenor/guides/quickstart) and put it in the `TENOR` variable in .env.
***
If you have any further questions regarding self-hosting, feel free to ask in the #support channel on the [esmBot Support server](https://esmbot.net/support).