Description
This project currently runs a bunch of RAPIDS notebooks, collects their logs in a few tarballs, and uploads those tarballs to downloads.rapids.ai.
See paths like the following at downloads.rapids.ai:
ci/docker/pull-request/{pr_number}/{commit_sha}/test_notebooks_output_*.tar.gz
That comes from here:
|
rapids-upload-to-s3 test_notebooks_output_${{ inputs.ARCH }}_cuda${{ inputs.CUDA_VER }}_py${{ inputs.PYTHON_VER }}_${{ inputs.GPU }}-${{ inputs.DRIVER }}.tar.gz /home/rapids/notebooks_output |
Those uploads should either be removed (if they are determined to be unused) or switched to using the GitHub Actions artifact store.
Benefits of this work
- another step towards removing
downloads.rapids.ai completely
- would allow contributors who don't work at NVIDIA to access those test result outputs from CI
Acceptance Criteria
- this repo does not depend on
downloads.rapids.ai in any way
- docs accurately reflect artifact-handling in this repo
Approach
Try using actions/upload-artifact in test workflows, similar to this:
https://github.com/rapidsai/shared-workflows/blob/d87850d9ccd0c6b1c610a0bf533e1acf1731f844/.github/workflows/wheels-build.yaml#L272-L277
Notes
N/A
Description
This project currently runs a bunch of RAPIDS notebooks, collects their logs in a few tarballs, and uploads those tarballs to
downloads.rapids.ai.See paths like the following at
downloads.rapids.ai:That comes from here:
docker/.github/workflows/test-notebooks.yml
Line 108 in bebddd8
Those uploads should either be removed (if they are determined to be unused) or switched to using the GitHub Actions artifact store.
Benefits of this work
downloads.rapids.aicompletelyAcceptance Criteria
downloads.rapids.aiin any wayApproach
Try using
actions/upload-artifactin test workflows, similar to this:https://github.com/rapidsai/shared-workflows/blob/d87850d9ccd0c6b1c610a0bf533e1acf1731f844/.github/workflows/wheels-build.yaml#L272-L277
Notes
N/A