Browser Terminal
Open a root shell on a running machine from the console, without an SSH client, and understand which providers support it and why the others don't.
Every Aquanode machine runs a small agent on the box. On providers that give the machine a directly reachable address, the console can attach a terminal straight to that agent, so you get a root shell on the machine from a browser tab, with no SSH client involved.
Opening a terminal
Open a running deployment from Deployments.
If a terminal is available, the deployment shows Connection Available next to an Open Terminal control, alongside the usual ssh command.
Click the Terminal tab. The session connects and drops you at a root prompt on the machine.
It is a real shell, not a restricted console. nvidia-smi, apt, pip, editing files, tailing a training log, attaching to tmux, all of it behaves exactly as it does over SSH, because it is the same shell.
Provider support
The terminal proxy binds a fixed port on the machine's own network interface. That means the console can only reach it when the provider gave your box a directly dialable address.
| Provider | Browser terminal |
|---|---|
| Verda | Supported |
| Hot Aisle | Supported |
| Massed Compute | Supported |
| RunPod | Supported |
| VoltagePark | Supported |
| Vultr | Supported |
| Akash | Not available (NAT / forwarded ports) |
| Hyperstack | Not available (NAT / forwarded ports) |
| SimplePod (docker pool) | Not available (NAT / forwarded ports) |
| Vast.ai | Not available (NAT / forwarded ports) |
On the providers in the lower group, your machine is reached through a per-machine forwarded port rather than its own address. There is no guarantee the terminal's port is forwarded too, so Aquanode refuses to guess rather than hand you a URL that silently connects to something else.
When a terminal is unavailable, the deployment records the reason and the
console shows it, rather than presenting a tab that fails. Use the ssh
command shown on the deployment page instead. SSH is available on every
machine regardless of provider.
Terminal or SSH?
Use SSH for real work: long sessions, port forwarding, scp and rsync, editors over Remote-SSH. The aq CLI makes that a single command.
The browser terminal is for the moments when opening a client is the annoying part:
- checking whether the GPU is actually busy before you conclude a job is stuck
- reading why a startup script did not do what you expected
- fixing one line in a config from a machine that does not have your key on it
- letting a teammate with console access look at a box without first adding their public key and redeploying
Requirements
- The deployment must be running. A paused or closed deployment has no live agent to attach to.
- The provider must assign the machine a directly dialable address, see the table above.
- No SSH key is needed for the browser terminal itself, though SSH keys are still required to create a machine.