VALDI: Start and connect to a Direct VM - storj/secure-stable-diffusion GitHub Wiki

Prerequisites

Go to the Direct services

Log in to the VALDI dashboard and select the "Direct" button from the VALDI Services section.

image

Select a GPU

While under the Virtual Machines section, look at the available VMs. Select a GPU with 16 GB RAM or more.

image

Configure the VM

Recommended specification:

  • Number of vCPUs: 8
  • Amount of RAM (GB): 16
  • Number of GPUs: 1
  • Local Storage Size (GB): 50 or more
  • Operating System: Ubuntu
  • Select the checkboxes for all ports

image

Launch the VM

Click the Launch Virtual Machine button.

Review the price on the confirmation dialog and click the Launch Virtual Machine button again.

image

Copy the connect info for the VM to a safe place. You will need this info to connect to the VM in the next step. The first line is the SSH command to execute in the Terminal. The second line is the password to authenticate with.

image

Click the Okay button. You will be redirected to the Your Session section.

Connect to the VM

The Your Session section should show the VM status as Running. Click it.

image

This opens the VM info. Look at the SSH Port field. If it displays the SSH port number (e.g. 10001), you can continue with the next step and connect to the VM.

If the SSH port number is not displayed yet, the SSH agent is still starting. Wait a little bit and refresh the page until the SSH port number is displayed.

image

Open the command terminal of your operating system and execute the command from the previous step to connect to the VM via SSH.

ssh -p <ssh-port> user@<ip-address>

Replace <ssh-port> and <ip-address> with the actual SSH port and IP address of your VM.

Respond with yes if prompted to accept a key fingerprint.

When prompted, enter the password (again, you should have it from the previous step).

You should see a Welcome to your VALDI Direct server. message and system information.

image

You are now connected to the VM.

Next step

Launch the Docker container in inference mode.

⚠️ **GitHub.com Fallback** ⚠️