Create rockcraft projects for pytorch full CPU image

[DnPlas]

Why it needs to get done

The DSS project allows users to select the image of the Notebook server to be used, some of which are built and distributed by this team.

What needs to get done

1. Create a rockcraft project for the pytorch full Notebook Server image. Use this Dockerfile for reference.
2. Add sanity tests to the rock.

When is the task considered done

When the pytroch full CPU image is published in the CKF team Dockerhub repository.

[syncronize-issues-to-jira[bot]]

Thank you for reporting us your feedback!

The internal ticket has been created: https://warthogs.atlassian.net/browse/KF-5462.

This message was autogenerated

[misohu]

Today I was working on this looks like we already have this rock created for kubeflow here. After discussion on daily we decided to use those rocks form kubeflow-rocks repo. On top of these we will install mlflow library. At the time of planning we forgot about those rocks.

[misohu]

Checking the code in here today. I found out that the upstream branch 1.8-branch has changed completely so we can no longer build the rocks the same way as we did originally. Key differences:

• each sub base image (jupyter, jupyter-pytorch) is not installing key requirements from requirements.txts but rather inside the docker file example
• names of the files has changed e.g. ther is no longer cpu-requirements.txt in jupyter-pytorch folder.
• upstream uses mamba to install dependancies to support arm64

because of this we need to rewrite each rockcraft yaml.

[misohu]

After talking with @DnPlas we agreed on following:

• I will rewrite the pytorch cpu jupyter ROCK to match the upstream Dockerfile except the s6 updates which are not needed for this task. PR is here.
• I will open this issue.

[misohu]

During the testing of the new ROCK I was facing problems with trivy timing out. I openned this PR to increase the default timeout https://github.com/canonical/charmed-kubeflow-workflows/pull/51