Before you begin
You'll need a Linux server, a public domain name, and access to configure your network. One Linux host with a DNS name serves the technician portal, the join page, the browser console and the WebSocket relay. There is no build step.
Browsers won't open a WebSocket to an untrusted certificate, so make sure ports 80 and 443 are reachable for Let's Encrypt.
Choose your installation method
Debian or Ubuntu
Recommended. The installer sets up Node.js, PostgreSQL and certbot for you.
Learn moreDocker Compose + Caddy
Container-based, with automatic HTTPS.
Learn moreCustom Linux
Different distro or your own database.
Learn moreOption 1 — Debian or Ubuntu (recommended)
Download and install
Download the latest release, extract it, and run the installer from the extracted folder.
tar -xf secure-desk-*.zip cd rubix-remote sudo bash install.shThe installer configures Node.js, PostgreSQL and certbot, creates the database role (with a generated password), installs the systemd service and applies the schema on boot.
Get your setup token
Until setup is complete the relay sends every request to
/setupand only accepts a one-time token it prints to its log:sudo journalctl -u rubix-relay -e | grep -i 'setup token'Keep this token secure.
Complete your portal setup
Open
https://your-domain.com/setup, paste the token and create your admin account. The wizard also takes the listening port, the Let's Encrypt domain and contact address, and the public base URL used in emailed links when you're behind a proxy.Sign in to your portal
Open
https://your-domain.com/portaland sign in with your new administrator account. You're ready to activate your license and connect your first device.
The commands above follow the current release process. Confirm the archive filename and directory name against the package you download before running them.
Option 2 — Docker Compose + Caddy
For teams that would rather ship a container than manage a VM. Caddy obtains and renews the certificate automatically (HTTP-01/TLS-ALPN) and proxies WebSocket upgrades transparently. Certificates and configuration persist in named volumes.
Certificates persist in the caddy-data volume and relay configuration in relay-data. Build with the repository root as the context — the supplied docker-compose.yml already sets context: .. so the server can find the portal/, join/, agent-web/ and setup/ folders. To upgrade, rebuild and bring the stack back up.
Option 3 — Custom Linux installation
On a non-Debian VM, install Node.js 18+ yourself and point DATABASE_URL at your own PostgreSQL — this path doesn't provision a database for you.
Without PostgreSQL the relay starts against an in-memory store and loses every account, session and device on restart. That mode exists for developers only — it is not a deployment.
Using a reverse proxy
Set the public base URL in the setup wizard if you run behind a proxy. The portal builds join links from it — get it wrong and the links you email point somewhere users can't reach, which looks exactly like the relay being down.
Portal URLs
/and/join- The public join page. A user opens a link, or types the code they were given.
/portal- Technicians: sign in, create and share sessions, browse enrolled devices, open the console.
/console- The browser support console, normally opened from the portal with the session prefilled.
/setup- First-run wizard. Closes itself once the deployment is configured.
Updating Secure Desk
Re-run the same installer command. It re-syncs the code, re-applies the schema and restarts the service, preserving the database, generated secrets and configuration. Under Docker, rebuild and bring the stack back up. Your portal can also tell you when a newer signed build is available (Settings → License → Check for updates).
Windows clients update themselves from your relay — see Licensing & updates.
Troubleshooting
- Join links don't work for users — check the public base URL setting if you're behind a proxy.
- Browser won't connect to the relay — make sure the certificate is trusted and ports 80/443 are reachable.
- Locked out after a TLS change — run the installer with
--reset-httpto revert to plain HTTP on port 80. - Accounts disappear after restart — the relay is running without PostgreSQL; configure
DATABASE_URL.
Installer options
| Flag | What it does |
|---|---|
--port N | Listen on a different port. |
--uninstall | Remove the service. Keeps the database and data. |
--factory-reset | Wipe accounts, settings and data; keep the application. Back to /setup. |
--purge | Destructive. Removes everything, database included. |
--reset-http | Recovery: revert to plain HTTP on port 80 when a TLS change has locked you out. |
Next: activate your license
An unlicensed deployment runs as free Starter — 1 technician, 1 active session and 3 enrolled Windows devices. Get your free license or read about licensing.
Something unclear or missing? Tell us and we'll improve this page.