Create a new workspace
To create a workspace, you need to select few options including workspace name, cluster, resource, image and few advanced settings.
Workspace name
Once you start to create a new workspace, the default workspace name will be randomly generated. Specify a good name to remember.
Cluster
You can choose between VESSL Cloud, which is a cluster managed by VESSL AI, or a custom cluster that you register yourself. (See configure organization cluster for more details.) VESSL Cloud is always hosted on a cloud vendor’s server, while the custom cluster can be hosted either on a cloud server or on an on-premise server.
We support the following integrations:
- AWS EKS
- GCP GKE
Resource
Choose the type of resource that the container will use. Select the resource among the dropdown option or specify the requirements manually.
Max runtime
Specify the max runtime for this workspace. After max runtime, workspace will be automatically stopped.
Image
You can choose the Docker image that the workspace container will use. There are two types of images: the Managed Image and the Custom Image. Select the Docker image type that you want to run on the workspace container.
For the Managed Image, you can simply select such an option, then the image managed by VESSL AI will be used in default. You can run Jupyter services on the managed image.
For the Managed Image, you can simply select such an option, then the image managed by VESSL AI will be used in default. You can run Jupyter services on the managed image.
You can use any docker images from Docker Hub or AWS ECR. To run a workspace with custom images, your custom images have to satisfy below requirements.
- Jupyterlab
- VESSL runs Jupyterlab and expose port
8888. Jupyter should be pre-installed in the container image. - sshd
- VESSL runs sshd and expose port
22as NodePort. sshd package should be pre-installed in the container image. - Persistent Volume Claim (PVC) mountable at
/root - VESSL mounts a PVC at
/rootto keep state across Pod restarts.
Public image
To pull images from the public Docker registry, you can simply pass the image URL as the below example.
Private image
To pull images from the private Docker registry or the private AWS ECR, you should integrate your credentials in organization settings first. Then check the private image checkbox and select the credentials you have just integrated. Below is an example of a private image from the AWS ECR.
Advanced Setting
Disk
You can specify the disk size to use in your container. This will be the request storage size of your PVC. The disk size cannot be changed once the workspace is created.
Port
You can customize port settings. By default, 8888 (jupyter) and 22 (ssh) are exposed.
Initialization (Init) script
Init script is a shell script that runs every time the workspace starts. Because /root is the only persistent directory, packages you installed outside the home directory may reset on stop and start. In this case, you can fill init script with install commands such as apt-get update && apt-get install ripgrep -y.