Skip to content
New issue

Have a question about this project? Sign up for a free GitHub account to open an issue and contact its maintainers and the community.

By clicking “Sign up for GitHub”, you agree to our terms of service and privacy statement. We’ll occasionally send you account related emails.

Already on GitHub? Sign in to your account

GHA Docker build flow for PR's #1883

Merged
merged 64 commits into from
Feb 14, 2023
Merged

GHA Docker build flow for PR's #1883

merged 64 commits into from
Feb 14, 2023

Conversation

bandish-shah
Copy link
Contributor

@bandish-shah bandish-shah commented Jan 12, 2023

What does this PR do?

Starting to move Docker builds to GHA. In this PR we:

  • Create the initial reusable Docker build workflow (docker-build.yaml)
    • Takes an image config YAML as input
    • Add's supporting script (gen_docker_matrix.yaml) to translate an image config YAML and translate it to a GHA matrix
    • Builds and optionally pushes images to a specified Docker repo
    • Requires repo DockerHub secrets
    • Reusable workflow still needs some work to be completely generic, cleanup will be handled in subsequent PR's as top-level pipelines are developed.
  • Create the top level PR Docker workflow (pr-docker.yaml)
    • Runs in PR's whenever repo Docker or Docker build CI files are touched
  • Updates image config generation script, adds IMAGE_NAME parameter
  • Upgrades flash-attn to 0.2.8
  • Testing for this PR:
    • Docker builds pass in GHA
    • Confirm images are pushed successfully to staging repo
    • Confirm image cache's are generated based on IMAGE_NAME
    • Confirm image cache works by re-triggering workflow after initial successful build, confirm speed-up
    • Manually verify pytest test suite works on Torch 1.13 images
  • Outside the scope of this PR:
    • We don't confirm that the pytest test suite works with all new images, this will be tested in a follow-up PR that merges the Docker build with the pytest testing workflows
    • Releasing Docker images to prod DockerHub repo's: mosaicml/pytorch, mosaicml/pytorch_vision, mosaicml/composer

What issue(s) does this change relate to?

Fixes CO-1064
Fixes CO-1606
Fixes CO-1762

To Do's Before Submitting

  • Ask IT to create a mosaicml/ci-staging repo, migrate off of mosaicml/jenkins-staging
  • Disable Jenkins Docker CI flows

Before submitting

  • Have you read the contributor guidelines?
  • Did you run the tests locally to make sure they pass?
  • Did you run pre-commit on your change? (see the pre-commit section of prerequisites)

@bandish-shah bandish-shah marked this pull request as ready for review January 19, 2023 17:50
@bandish-shah bandish-shah requested a review from a team as a code owner January 19, 2023 17:50
@bandish-shah bandish-shah changed the title GHA Docker Build Flow GHA Docker build flow for PR's Jan 19, 2023
Copy link
Contributor

@dakinggg dakinggg left a comment

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

LGTM, proof is in the pudding

@mvpatel2000
Copy link
Contributor

Holding review until EFA and GPU test lands

@bandish-shah bandish-shah merged commit fa83d50 into dev Feb 14, 2023
@bandish-shah bandish-shah deleted the bandish/docker_build branch February 14, 2023 06:40
Sign up for free to join this conversation on GitHub. Already have an account? Sign in to comment
Labels
None yet
Projects
None yet
Development

Successfully merging this pull request may close these issues.

3 participants