This repository contains the core codebase, installer and associated tools for instantiating and managing a HelioCloud instance in AWS.
The HelioCloud instantiation process is implemented as an AWS CDK project that - when provided an instance configuration pulls in the necessary CDK Stack definitions and instantiates/updates a HelioCloud instance in a configured AWS account.
For Daskhub specific cost monitoring, see Daskhub Cost Monitoring with Kubecost.
For general HelioCloud cost information, see presentations/posters with cost related information:
- HelioCloud: Collaborative Computing and Cost Management via Commercial Cloud
- Pop-up Cloud-Based Jupyter Platforms for Coding Camps Using HelioCloud
Deploying a HelioCloud instance is a simple matter of ensuring your local and AWS environments support the installation, setting a few configuration options to fine tune your deployment to your needs, running the CDK application and finally doing a few quick checks to confirm your HelioCloud instance is operating correctly.
The following is performed automatically by pre-commit.
After completing and testing your changes, you will want to take some additional steps to maximize the potential that the pull or merge request you put together is accepted. We recommend you:
- Run
blackto format all of your changes in keeping with HelioCloud's code formatting conventions
black .- Run
pylintto conduct a static code analysis of the codebase and see if you introduced any errors, deviations from coding standards, etc.
pylint *Provided you get clean feedback from black & pylint and your tests pass, you should feel pretty comfortable any merge request you post would get rejected for not adhering to the codebase conventions.
Stacks may import/export values with other stacks. For instance, the Auth stack exports Cognito client info to the Daskhub/Portal stacks.
This creates a coupling that makes altering these tools difficult. If an update needs to be made to the cognito clients or any other exported service, the old service must remain to ensure dependency is maintained. Once the new service is deployed, the stack can be redeployed with the old service deleted.