https://github.com/f3d-app/lfs-data-cache-action

A github action to cache and recover LFS data
https://github.com/f3d-app/lfs-data-cache-action

Last synced: 12 days ago
JSON representation

A github action to cache and recover LFS data

• Host: GitHub
• URL: https://github.com/f3d-app/lfs-data-cache-action
• Owner: f3d-app
• License: bsd-3-clause
• Created: 2023-09-17T13:01:04.000Z (over 1 year ago)
• Default Branch: main
• Last Pushed: 2024-10-21T21:20:45.000Z (3 months ago)
• Last Synced: 2024-11-13T23:40:01.525Z (2 months ago)
• Language: CMake
• Size: 8.79 KB
• Stars: 2
• Watchers: 3
• Forks: 1
• Open Issues: 0
• Metadata Files:
  • Readme: README.md
  • License: LICENSE

Awesome Lists containing this project

README

        
# lfs-data-cache-action

A producer/consumer github action to cache and recover LFS data.

Its main use is to avoid cloning LFS data in CI to avoid

having to pay for LFS bandwidth because of CI needs.

It expects cmake to be available on the host.

The action can be used as a consumer or a producer, and must

provide the repository containing the LFS data to recover from.

It is possible to provide a specific SHA to produce.

If not provided, the last commit modyfing the LFS file will be produced.

Is has the following inputs:

 - `type`: should be `producer` or `consumer`, default to `producer`

 - `repository`: the git repository to produce LFS data from, default: ${{ github.repository }}

 - `lfs_sha`: The git sha to recover LFS data from, optional

 - `cache_postfix`: An postfix added to the cache name, to support multiple caches, default to `cache`

 - `target_directory`: A target directory to copy LFS data to

## Logic

Producer/Consumer first use the classic cache action to recover a cache named

`lfs-data-${{lfs_sha}}-${{cache_index}}`.

If Producer does not found it, it will clone the `repository` at `lfs_sha` commit

and upload the content as an artifact.

If Consumer does not found it, it will try to download a potential artifact

produced earlier by the Producer.

If it fails Consumer will clone the `repository` at `lfs_sha` commit.

Finally, Producer/Consumer will copy the LFS data only using cmake to the `target_directory`

## Usage

In an first job, use the `producer` action, which output the LFS sha that will be produced

In a second job, usually a matrix job, depending on the first,

recover the LFS sha from first job and use the `consumer` action.

```

jobs:

#----------------------------------------------------------------------------

# Cache LFS: Checkout LFS data and update the cache to limit LFS bandwidth

#----------------------------------------------------------------------------

  cache_lfs:

    runs-on: ubuntu-latest

    name: Update LFS data cache

    outputs:

      lfs_sha: ${{ steps.lfs_sha_recover.outputs.lfs_sha }}

    steps:

    # Checkout your repository WITHOUT LFS

    - name: Checkout

      uses: actions/checkout@v3

      with:

        path: 'source'

        fetch-depth: 1

        lfs: false

    # Use producer action to recover the LFS data and upload it as cache/artifacts

    - name: Cache LFS Data

      id: lfs_sha_recover

      uses: f3d-app/lfs-data-cache-action@v1

#----------------------------------------------------------------------------

# Actual CI: Recover LFS data first

#----------------------------------------------------------------------------

  recover_lfs:

    needs: cache_lfs

    # Checkout your repository WITHOUT LFS

    - name: Checkout

      uses: actions/checkout@v3

      with:

        path: 'source'

        fetch-depth: 0

        lfs: false

    - name: Recover LFS Data

      uses: f3d-app/lfs-data-cache-action@v1

      with:

        workflow_label: 'consumer'

        lfs_sha: ${{ needs.cache_lfs.outputs.lfs_sha}}

```