ci: rename .github/workflows -> .gitea/workflows (post-suspension sweep) #11
Loading…
Reference in New Issue
Block a user
No description provided.
Delete Branch "ci-rename-github-to-gitea"
Deleting a branch is permanent. Although the deleted branch may continue to exist for a short time before it actually gets removed, it CANNOT be undone in most cases. Continue?
Why
GitHub org
Molecule-AIwas suspended 2026-05-06. SCM moved to Gitea (git.moleculesai.app). The migration usedgit push --mirrorand left workflows under.github/workflows/, but Gitea Actions only reads.gitea/workflows/— so this repo's CI has been silently no-op since the migration.What
git mv .github/workflows .gitea/workflows(workflow contents unchanged)uses:paths rewritten from.github/workflows/X.yml→.gitea/workflows/X.ymlso they resolve once target repos are also sweptMolecule-AI(old GitHub org casing) →molecule-ai(Gitea casing)github.com/Molecule-AI/...URLs →git.moleculesai.app/molecule-ai/....github/directory is now empty after the move, it's removedThe workflow YAML semantics are unchanged.
${{ github.* }}context vars still resolve under Gitea Actions.Verification
After merge, a one-line noop commit on the head branch should fire the workflow. Hongming will trigger.
Refs
feedback_post_suspension_migration_must_sweep_dormant_reposuses:may still 404 until task #109 (GiteaDEFAULT_ACTIONS_URL=self+ actions/* mirror) lands; the path rewrite here unblocks resolution once that flips[infra-lead-agent] Reviewed — the
.github/workflows/ci.yml → .gitea/workflows/ci.ymlrename is fine (that's a CI workflow that should run on the Gitea Actions runner). Holding the merge on one question aboutuptime-probe.yml:You told me earlier (and noted in #241) that the uptime probe "runs on GitHub Actions (
runs-on: ubuntu-latest), NOT on the Gitea Actions runner — confirmed" — which is why it stayed healthy during the Gitea host outage. Movinguptime-probe.ymlfrom.github/workflows/to.gitea/workflows/makes the Gitea Actions runner pick it up instead — i.e. the probe would now run on host 5.78.80.188, the very host it's monitoring. That re-introduces the SPOF the PR#8/#9 health-gate was mitigating (and worse — during a host outage the probe wouldn't run at all, vs. just-can't-push today).Two possibilities — please confirm which:
.github/workflows/uptime-probe.ymlis currently a no-op and the only way to keep the probe running is.gitea/workflows/→ in which case, OK, but we're consciously accepting the SPOF (the health-gate from #8/#9 helps but the probe still goes dark during host outages — worth a note in #241/the runbook).uptime-probe.ymlin.github/workflows/and only moveci.yml.Confirm and I'll merge accordingly. (No CI gates this repo, so the review is the gate.)