Skip to content
rdrd

Linux agent

On Linux, rd-agent has no screen. It serves the terminal and the file browser: a shell on the server in your browser tab, over the same port 443 as everything else in rd. The server needs no SSH and no open ports; like the Mac and Windows agents, it dials out to the gateway.

The portal shows the server with Terminal and Files, but no Connect. Shells run as the user the agent runs as.

Requirements

Section titled “Requirements”
  • Any Linux on x86-64 or ARM64 (the release builds are static: no particular distribution or glibc). Agent releases from 1.11.0 on include them.
  • curl, and systemd for the user unit (without systemd, run rd-agent run from your own service manager).

1. Add the server in the portal

Section titled “1. Add the server in the portal”

Click Add device in the portal (Devices). Use a machine id such as box or the host name. The dialog shows the setup code, and under Linux server the whole install command with the code filled in.

2. Install

Section titled “2. Install”

On the server, as the user whose shell you want in the browser (not root):

Terminal window
curl -fsSL https://raw.githubusercontent.com/Thijmen/remoted/main/packaging/linux/install.sh | sh -s -- <setup code>

install.sh downloads the newest agent release for your architecture from GitHub, checks it against the release’s SHA256SUMS.txt, installs it as ~/.local/bin/rd-agent and runs rd-agent setup --code <setup code>. Setup writes ~/.config/rd-agent/config.toml (mode 0600, with the token), and installs, enables and starts a systemd user unit, ~/.config/systemd/user/rd-agent.service. There are no permissions to grant.

Without a setup code the script asks setup’s questions instead. RD_AGENT_VERSION=1.11.0 installs a given version; RD_INSTALL_DIR puts the binary elsewhere.

Updating: the Linux agent does not update itself yet. Run the same command without the setup code; it replaces the binary and restarts the agent:

Terminal window
curl -fsSL https://raw.githubusercontent.com/Thijmen/remoted/main/packaging/linux/install.sh | sh

Building it yourself instead: Rust (the version in rust-toolchain.toml), a C compiler and CMake (for Opus), then cargo build --release -p rd-agent, copy target/release/rd-agent to ~/.local/bin and run rd-agent setup.

User services stop when you log out. To keep the agent running without a login, and start it at boot:

Terminal window
sudo loginctl enable-linger "$USER"

3. Check

Section titled “3. Check”
Terminal window
rd-agent status
journalctl --user -u rd-agent -f

Logs also go to ~/.local/state/rd-agent/logs/ (one file a day, 7 kept).

The terminal on Linux

Section titled “The terminal on Linux”

The terminal works as on the other agents (Terminal): shells survive a dropped connection and the [terminal] settings apply. The default shell is your login shell ($SHELL, else the user database), started as a login shell with TERM=xterm-256color; set another with shell = "/usr/bin/fish" under [terminal] in the config file.

Remove

Section titled “Remove”
Terminal window
rd-agent uninstall          # stops and removes the systemd user unit
rm -r ~/.config/rd-agent ~/.local/state/rd-agent ~/.local/bin/rd-agent

Then remove the device in the portal.

Local development

Section titled “Local development”
Terminal window
rd-gateway --dev                                           # http://127.0.0.1:8080, agents on :8081
rd-agent run --gateway ws://127.0.0.1:8081 --machine box   # with RD_AGENT_TOKEN=rd1.box.dev

rd-agent run --terminal-only on macOS or Windows runs an agent the same way: no screen, only the terminal and files.