search

redhat-et / ray-on-the-edge-demo

A proof of concept for machine learning workflows using the Ray stack in the data center and at the edge.
Apache License 2.0
0 stars 2 forks source link

Make ray cluster control independent of image used #15

Closed MichaelClifford closed 2 years ago

MichaelClifford commented 2 years ago

This PR includes major changes to how we create, destroy and interact with Ray Clusters on OpenShift. This change is primarily reflected in the addition of cluster_template.yaml, ray_cluster_control.py, and example.env. Instead of relying on a predefined cluster yaml that spins up and down based on a single-user-profile in JupyterHub, we can now use a .env file to configure our ray cluster from within our notebook environment, apply these configurations to the cluster_template.yaml, and then start and stop our clusters by calling functions from ray_cluster_control.py.

This change should make Ray usage more flexible by giving users greater control over their clusters and making its use independent of any specific notebook image (on Jupyterhub or KFNBC).

This PR also updated the E2E demo notebooks to reflect the new way in which we control the ray clusters.

This PR also updates the requirements.txt to make sure all the demos work correctly.

review-notebook-app[bot] commented 2 years ago

Check out this pull request on  ReviewNB

See visual diffs & provide feedback on Jupyter Notebooks.

Powered by ReviewNB