UserVM Handbook: Difference between revisions
No edit summary |
(add KillMode=mixed to .service template) |
||
(One intermediate revision by the same user not shown) | |||
Line 91: | Line 91: | ||
qemuArgs = "qemu-system-x86_64 -M q35,usb=on,acpi=on,hpet=off -cpu host -accel whpx -m 2G -smp cores=2 -device usb-tablet -nic none -hda /srv/collabvm/images/vm1.qcow2" |
qemuArgs = "qemu-system-x86_64 -M q35,usb=on,acpi=on,hpet=off -cpu host -accel whpx -m 2G -smp cores=2 -device usb-tablet -nic none -hda /srv/collabvm/images/vm1.qcow2" |
||
</nowiki>}} |
</nowiki>}} |
||
<!-- expansion on this is tbd --> |
|||
Additionally, it is possible to use [https://github.com/modeco80/lilyvm LilyVM] to help with building more complex VM configurations. |
|||
== Running your VM == |
== Running your VM == |
||
Line 132: | Line 135: | ||
[Service] |
[Service] |
||
⚫ | |||
Type=simple |
Type=simple |
||
User=collabvm |
User=collabvm |
||
Group=collabvm |
Group=collabvm |
||
KillMode=mixed |
|||
⚫ | |||
RestartSec=5 |
|||
# Make sure to change the following two lines according to where you put your server. |
# Make sure to change the following two lines according to where you put your server. |
||
# If you have multiple VMs, |
# If you have multiple VMs, it's possible to make your service file a template unit file (by making the file name, for example, "[email protected]"), and use %i in WorkingDirectory |
||
# allowing you to use the same server for all your VMs. |
# to automatically set WorkingDirectory to a different directory for each VM, allowing you to use the same server for all your VMs. |
||
WorkingDirectory=/srv/collabvm/collabvm-1.2.ts/ |
WorkingDirectory=/srv/collabvm/collabvm-1.2.ts/ |
||
ExecStart=/bin/node /srv/collabvm/collabvm-1.2.ts/cvmts/dist/index.js |
ExecStart=/bin/node /srv/collabvm/collabvm-1.2.ts/cvmts/dist/index.js |
||
# Tell systemd that we manage our own cgroup hierarchy, and delegate |
|||
# all controllers that are either implicitly or explicitly enabled. |
|||
# |
|||
# This is used for resource limits (in your VM's config.toml). |
|||
# Can be omitted if you are not using it. (It's probably a good idea to however!) |
|||
Delegate=yes |
|||
# Hardening |
|||
PrivateTmp=yes |
|||
NoNewPrivileges=true |
|||
RestrictNamespaces=uts ipc pid user cgroup |
|||
ProtectKernelTunables=yes |
|||
ProtectKernelModules=yes |
|||
PrivateDevices=no |
|||
RestrictSUIDSGID=true |
|||
[Install] |
[Install] |
Latest revision as of 16:08, 2 November 2024
Welcome to the official guide on setting up a UserVM.
The Rules
First, the boring part. We ask that all hosts review and follow the UserVM Hosting Rules.
Prerequisites
You'll need:
- A machine with decent specs (8GB of RAM and a modern CPU, probably)
- A Linux distribution; You can pick any mainstream distro, for the purposes of this guide I recommend either Debian or Arch, or their OpenRC counterparts if you prefer OpenRC. Yes, Ubuntu will work, it's terrible though
- If you REALLY want to run CollabVM on Windows, there is an unofficial and unsupported guide for that at UserVM Handbook/Windows
- A decently fast network that allows you to forward a port. We will not accept UserVMs behind services like ngrok. Cloudflare tunnels are fine. You must also have a URL that stays persistent. If your IP is dynamic, you can use services like NOIP or setup a script to auto-update your domain using cloudflare.
- Basic knowledge of how computers and Linux systems work. We aren't going to hold your hand, you need to be comfortable with a command line
- IF YOU DO NOT UNDERSTAND HOW TO FOLLOW THE INSTRUCTIONS IN THIS GUIDE, DO NOT HOST PUBLIC INTERNET CONNECTED VMS ON THE INTERNET THAT ANYONE CAN ACCESS
- A few hours
Compiling the server
Install Dependencies
First up make sure you have git, nasm, cmake, and a C(++) compiler installed
$ sudo pacman --needed --noconfirm -Sy git nasm base-devel cmake # Arch
$ sudo apt-get install -y git nasm build-essential cmake # Debian
Next, we need to install Node.js. First, we'll install node. On arch, you can just run the following command:
$ sudo pacman --needed -S npm nodejs
On Debian, the packaged node version is too old to run CollabVM, so we'll add the nodesource repository.
$ sudo apt-get install -y curl
$ curl -fsSL https://deb.nodesource.com/setup_21.x | sudo bash -
$ sudo apt-get install nodejs -y
Enable Corepack:
$ sudo corepack enable
Prepare the server
Now let's get the server ready. First, we'll create a dedicated CollabVM user to run the server from /srv/collabvm
:
$ sudo useradd -rmd /srv/collabvm collabvm
$ sudo usermod -aG kvm collabvm # Give the CollabVM user permission to use KVM hypervision
Now, we can shell in as the CollabVM user. For the remainder of this guide, any line that starts with (collabvm) $
indicates that this should be run as the collabvm
user.
$ sudo -iu collabvm
(collabvm) $ pwd # This should output /srv/collabvm
Install the Rustup toolchain for the CollabVM user:
(collabvm) $ curl --proto '=https' --tlsv1.2 -sSf https://sh.rustup.rs | sh
# Restart the shell to apply changes
(collabvm) $ exit
$ sudo -iu collabvm
Now, we can clone the CollabVM Server source code:
(collabvm) $ git clone https://github.com/computernewb/collabvm-1.2.ts.git /srv/collabvm/collabvm-1.2.ts --depth 1 --recursive
(collabvm) $ cd /srv/collabvm/collabvm-1.2.ts
Then install dependencies
(collabvm) $ yarn
Finally, build the server
(collabvm) $ yarn build
Set up your VM
Now is a good time to get your VM set up. Currently, the only supported hypervisor is QEMU. We have many guides on this wiki for setting up different OSes in QEMU, check them out here. Here are some ideas to make your VM interesting:
- A funny wallpaper
- Development software (Visual Studio, etc.)
- Some games
- Some harmless malware (for the love of god no GDI rapists)
We recommend setting up your VM as the collabvm user to make sure permissions are set correctly, but this is not a requirement.
Setting up a Virtual Network
QEMU's user-mode networking used by default isn't very customizable and lacks the ability to block certain abuse vectors. For this reason we very strongly recommend setting up a Virtual Network using the CollabNet Guide. Depending on the full situation we may refuse to add VMs that use QEMU user-mode networking.
It's also VERY important that mail ports are BLOCKED on your VM (the CollabNet guide config includes this). If you do not block them, your IP effectively becomes an open relay which will very likely get you suspended by your ISP or hosting provider. We will not add VMs with accessible mail ports
Configuration
Now we need to fill out the config file for your VM. Copy config.example.toml to config.toml, and open it in an editor. It is well commented so each value should be self-explanatory. If you have questions, feel free to ask in our Discord server.
QEMU Args
Example of a production QEMU command:
qemuArgs = "qemu-system-x86_64 -M q35,usb=on,acpi=on,hpet=off -cpu host -accel whpx -m 2G -smp cores=2 -device usb-tablet -nic none -hda /srv/collabvm/images/vm1.qcow2"
Additionally, it is possible to use LilyVM to help with building more complex VM configurations.
Running your VM
Now that everything is set up, you can bring your VM online. To run the server right from your terminal, run the following command:
(collabvm) $ yarn serve
Or alternatively, to run it directly:
(collabvm) $ node cvmts/dist/index.js
Running a local webapp
Before you put your VM on the UserVM roster, you'll probably want to test it out for yourself. For that, we'll throw up a test webapp. Start by cloning the source:
$ git clone https://github.com/computernewb/collab-vm-1.2-webapp.git --recursive
$ cd collab-vm-1.2-webapp
Then, copy config.example.json
to config.json
, and replace ServerAddresses with your server address:
"ServerAddresses": [
"ws://127.0.0.1:6004", # If you're not using proxying
"wss://example.com/collab-vm/vm1" # If you are using proxying. Remove one of these lines.
],
Now you can build the webapp, and serve it:
$ yarn
$ yarn build
$ yarn serve
This will run the webapp at 127.0.0.1:1234
, which you can navigate to in your browser. If all went well, your VM should show up. If not, and you don't know why, join our discord and ask for help there!
Setting up a service
While it's useful and convenient to run your VM from the console while debugging, we strongly recommend you set it up as a service once you're ready to leave it on for extended periods of time. This is done differently depending on what init system your distro uses (Probably systemd, if you're not sure)
Systemd
To make your VM a systemd service, you can put the following into /etc/systemd/system/collabvm.service
[Unit]
Description=CollabVM
[Service]
Type=simple
User=collabvm
Group=collabvm
KillMode=mixed
Restart=always
RestartSec=5
# Make sure to change the following two lines according to where you put your server.
# If you have multiple VMs, it's possible to make your service file a template unit file (by making the file name, for example, "[email protected]"), and use %i in WorkingDirectory
# to automatically set WorkingDirectory to a different directory for each VM, allowing you to use the same server for all your VMs.
WorkingDirectory=/srv/collabvm/collabvm-1.2.ts/
ExecStart=/bin/node /srv/collabvm/collabvm-1.2.ts/cvmts/dist/index.js
# Tell systemd that we manage our own cgroup hierarchy, and delegate
# all controllers that are either implicitly or explicitly enabled.
#
# This is used for resource limits (in your VM's config.toml).
# Can be omitted if you are not using it. (It's probably a good idea to however!)
Delegate=yes
# Hardening
PrivateTmp=yes
NoNewPrivileges=true
RestrictNamespaces=uts ipc pid user cgroup
ProtectKernelTunables=yes
ProtectKernelModules=yes
PrivateDevices=no
RestrictSUIDSGID=true
[Install]
WantedBy=multi-user.target
Reload the daemon cache:
$ sudo systemctl daemon-reload
Then you can start your VM with:
$ sudo systemctl start collabvm
And make it automatically run on startup with:
$ sudo systemctl enable collabvm
OpenRC
Put the following into /etc/init.d/collabvm to make your VM an OpenRC service (change filename as appropriate):
#!/sbin/openrc-run
supervisor="supervise-daemon"
name="collabvm"
command="/bin/node"
command_args="/srv/collabvm/collabvm-1.2.ts/cvmts/dist/index.js"
# If you have multiple VMs, you can change --chdir to a different directory on each VM, to use different config files on the same server
supervise_daemon_args="--user collabvm --group collabvm --chdir /srv/collabvm/collabvm-1.2.ts --stdout /srv/collabvm/out.log --stderr /srv/collabvm/error.log"
Make it executable:
$ sudo chmod +x /etc/init.d/collabvm
Now you can start your VM with:
$ sudo rc-service collabvm start
And make it run on startup with:
$ sudo rc-update add collabvm
Setting up reverse proxying
This is REQUIRED for UserVM as, for technical reasons, only TLS-equipped WebSockets can be accepted
We strongly recommend you proxy your UserVM behind Nginx, to provide additional security and allow things like TLS. It also makes your VM look a lot cleaner, allowing people to access it on your main HTTP(s) port and on a subdirectory, like https://example.com/collab-vm/
rather than http://example.com:6004
. Here's a brief description of how to set that up on the Nginx side. This assumes you already have your site set up with Nginx, and if not there are numerous guides for that around the internet.
First, you'll want to save wsproxy_params to your Nginx directory, which enables WebSocket proxying. Here's a one-liner to do that:
$ sudo curl https://computernewb.com/~elijah/wsproxy_params -o /etc/nginx/wsproxy_params
Next, you can add the following to your Nginx server block:
location /collab-vm/vm1 {
include wsproxy_params;
proxy_pass http://127.0.0.1:6004/; # Replace 6004 if you changed the HTTP port in the config file.
}
If you get an error about connection_upgrade
, edit /etc/nginx/nginx.conf
and add the following to your http block:
map $http_upgrade $connection_upgrade {
default upgrade;
'' close;
}
If you have multiple VMs running, you can have them all proxied like so:
location /collab-vm/vm1 {
include wsproxy_params;
proxy_pass http://127.0.0.1:6004/;
}
location /collab-vm/vm2 {
include wsproxy_params;
proxy_pass http://127.0.0.1:6005/;
}
# ...etc
Permanently host the webapp
If you want to host the webapp on your website, you can build it as follows:
$ yarn build
Then, copy the contents of the dist
directory to your website. For example, if your webroot is at /var/www/example.com
, and you want your webapp at example.com/collab-vm/:
$ cp -r dist/. /var/www/example.com/collab-vm/
The webapp should now be accessible at your website.
Logging in as an admin (or mod)
Logging in is very simple. Just join the VM, and double click your username. Enter your admin or mod password into the prompt, and you should be authenticated and able to use staff actions.
Getting your UserVM on the roster
Now you can have your UserVM put on the roster! Join our Discord, and create a post in #support with the uservm roster update
tag, including your VM's WebSocket URL. You can ping me (@elijahr.dev) if you'd like.