Closed PepperMoJ closed 1 month ago
The GitHub Actions Importer seems to be the most promising way of analysing, forecasting and automating the conversion of our CircleCI pipelines.
A report from the GitHub Actions Importer tool with an audit of our migration potential:
Summary for CircleCI instance
Total: 505
Supported: 412 (81%)
Unsupported: 93 (18%)
Total: 18938
Known: 15308 (80%)
Unknown: 3154 (16%)
Unsupported: 476 (2%)
Actions: 17497
Total: 1934
Known: 1446 (74%)
Unsupported: 488 (25%)
Actions: 1446
Total: 0
Actions: 10675
Total: 328
Known: 315 (96%)
Unsupported: 13 (3%)
Actions: 30373
Total: 1230
Secrets: 1230
${{ secrets.AWS_ACCESS_KEY_ID }}
: 199${{ secrets.AWS_SECRET_ACCESS_KEY }}
: 199${{ secrets.CIRCLE_CI_SSH_KEY }}
: 140${{ secrets.CIRCLE_CI_KNOWN_HOSTS }}
: 140${{ secrets.AWS_ACCESS_KEY_ID_ENG }}
: 105${{ secrets.AWS_SECRET_ACCESS_KEY_ENG }}
: 105${{ secrets.AWS_KEY_ID }}
: 167${{ secrets.AWS_SECRET_KEY }}
: 167${{ secrets.AWS_ACCESS_KEY_ID_PREPROD }}
: 3${{ secrets.AWS_SECRET_ACCESS_KEY_PREPROD }}
: 3${{ secrets.AWS_ACCESS_KEY_ID_PROD }}
: 1${{ secrets.AWS_SECRET_ACCESS_KEY_PROD }}
: 1Here's the output for the moj-analytical-services
organisation on CircleCI:
Summary for CircleCI instance
Total: 68
Supported: 1 (1%)
Unsupported: 67 (98%)
Total: 4
Known: 4 (100%)
Actions: 4
Total: 1
Known: 1 (100%)
Actions: 1
Total: 0
Total: 1
Known: 1 (100%)
Actions: 1
Most computationally intensive projects can be seen on the usage overview.
Background
We are currently using a large amount of credits across our CircleCI projects. This ticket is to investigate
Questions / Assumptions
Tasklist
What hypothesis do we want to test?/What do we want to learn?
Can we automate the conversion of CircleCI workflows to GitHub Actions?
Definition of done