Environment Configs

Sessions has a variety of fully managed Jupiter environments for developing VSCode notebooks, python files, and Jupyter notebooks.

To spin up a Sessions instance, go to the Sessions tab on the dashboard and start a new session.

Below is a list of pre-built for a quick start.

Display Name	Image Name	Best For	Key Packages
Minibasic	minibasic	basic data processing	numpy, pandas
Basic	basic	data science	numpy, scipy, scikit-learn, Dask
Deep learning	dl	deep learning	TensorFlow, Pytorch, MXNet, transformers
GPU	gpu	machine learning with GPU	Rapids TensorFlow Pytorch
Triton model serving	triton	model serving with triton	tritonclient
Ray Spark	rayspark	data engineering with spark on ray	Ray raydp Spark
Ray	ray	distributed training, hyperparameter tuning	Ray
Crypto	crypto	crypto trading and DApps	ccxt web3
Climate	climate	large scale climate data processing	xclim Pangeo xarray
SAS on Jupyter	sas	run sas code in Jupyter and integrate with python	saspy
R on Jupyter	rstat	run R code in Jupyter and integrate with python	rstat
Rust	rust	building DApps	rust, cargo, SOL SDK
Custom	custom	use your own Docker image	-

---

Environment Config details

To check the configurations for each EC, navigate to the Environment Config tab and click on the card you'd like to see the details for.

Check installed packages

To check what packages are installed in an EC, you can do so from Sessions.

1. Create a new session and choose the session type for which you'd like to check the packages for.

2. Open up the browser Session and open a Jupyter notebook.

3. Type in pip freeze to see pip packages installed and conda list for conda packages installed.

---

Create an EC

Prerequisities

These are a few requirements when creating an EC:

• An image that is accessible to Shakudo
• Volumes and volume mounts to exist
• Appropriate node pools or resources

Steps to create an EC

You can create a new EC to use in your Sessions, Jobs, or Services. Follow the steps below to do so:

1. Navigate to the Environment Config tab and click on the + button to create a new EC

2. Customize the details. For details on each field, see the section below for podspec configurations

3. Click CREATE

4. To view and use the new EC you must press ⌘ + R to hard refresh the page.

---

EC configurations

Name

This is the name that will show up when choosing which EC to use in Sessions, Jobs, and Services.

EC name

This name acts as a unique ID for the EC.

Description

Give the EC a one liner description for when someone should pick it. This description appears in the EC page as well as in the Session Type selection.

Image URL

You can choose to bring your own docker image to build your custom EC with. Paste in the link to the image in this field.

Resources

You can specify the CPU request & limit as well as the memory request & limit.

If you want GPUs make sure to specify GPUs that are available to your cluster. If you're unsure please contact us and we can provide you the details.

Volumes & volume mounts

Specify your Volumes and Volume Mounts to attach by clicking on the + button. For more information see the Kubernetes guide on Volumes.

Environment variables

Use this field to specify the name and value for environment variables.

---

Edit an EC

You can edit details of an EC by clicking on the pencil icon from the EC tab. By default you can only edit the custom created ECs (not the default ECs created by Shakudo)

When an EC is edited and saved, newly created jobs and sessions will use the updated version. Any jobs or sessions currently in progress will use the old version. Any services in progress using the edited EC will be restarted to update.