potiuk opened a new issue, #1731: URL: https://github.com/apache/airflow-site/issues/1731
In `apache/airflow-site-archive`, `github-to-s3.yml` and `s3-to-github.yml` are the only workflows still on `ubuntu-latest` (see apache/airflow-site-archive#49, which moved the rest to `ubuntu-slim`). They can't use `ubuntu-slim`, and they are a poor fit for standard GitHub-hosted runners too: - **Disk:** a full sync checks out the docs archive (~900k files, tens of GB) into `/mnt`. We already have to delete preinstalled toolchains to make room. - **Runtime:** full syncs and back-reference generation run for a long time and use GitHub-hosted runner minutes. - **Setup:** every run installs the AWS CLI with `sudo`, plus `uv` and breeze (for `s3-to-github.yml`). Proposal: move both workflows to self-hosted runners with enough persistent disk and the AWS CLI preinstalled. Things to work out: - [ ] Which self-hosted runner pool/labels to use, and whether it is allowed to run `workflow_dispatch` jobs that use `DOCS_AWS_*` secrets - [ ] Clean the workspace between runs so no state or credentials leak between jobs (or use ephemeral runners) - [ ] Optionally keep a persistent clone of the archive on the runner to skip the full checkout each time - [ ] Remove the "Remove some stuff we don't need" and AWS CLI install steps once the runner image provides what's needed 🤖 Generated with [Claude Code](https://claude.com/claude-code) -- This is an automated message from the Apache Git Service. To respond to the message, please log on to GitHub and use the URL above to go to the specific comment. To unsubscribe, e-mail: [email protected] For queries about this service, please contact Infrastructure at: [email protected]
