ci: consolidate release workflow and trim GHCR tags
test / test (push) Has been cancelled

Run release.yml directly on release:published instead of a dispatcher
workflow. Remove per-arch tags from GHCR after the manifest merge so only
X.Y.Z remains visible to operators.

Co-authored-by: Cursor <cursoragent@cursor.com>
This commit is contained in:
2026-08-14 21:44:33 +03:00
parent e17b1680bd
commit 5601f73622
3 changed files with 48 additions and 96 deletions
+14 -13
View File
@@ -205,9 +205,8 @@ the binaries so they cannot drift apart.
local-trial image references) in the **same** release commit.
2. Create and push git tag `vX.Y.Z` on that commit.
3. Publish the GitHub Release for `vX.Y.Z` (not a draft).
4. Workflow [release-on-publish.yml](../.github/workflows/release-on-publish.yml)
starts [release.yml](../.github/workflows/release.yml) with that version;
the build checks out tag `vX.Y.Z` (not `main` HEAD).
4. Workflow [release.yml](../.github/workflows/release.yml) builds, e2e-gates,
and publishes `ghcr.io/mixeme/selfpost:X.Y.Z` (checks out tag `vX.Y.Z`).
**GitHub Release vs GHCR.** The public [Releases](https://github.com/mixeme/selfpost/releases)
page lists only **published** releases. A draft is visible to maintainers only —
@@ -307,29 +306,31 @@ Workflows in [.github/workflows/](../.github/workflows/). What each job runs —
### `release.yml` — published GitHub Release, or `workflow_dispatch` with SemVer
Publishing a GitHub Release runs [release-on-publish.yml](../.github/workflows/release-on-publish.yml),
which dispatches `release.yml` with the version parsed from the release tag.
You can also run `release.yml` manually via `workflow_dispatch` and an
explicit `X.Y.Z` input. A bare git tag push does not run either workflow.
The build always checks out `vX.Y.Z`, not `main` HEAD.
Publishing a GitHub Release runs `release.yml` directly (`release: published`,
same pattern as gosentry / imap-scrub). You can also run it manually via
`workflow_dispatch` with an explicit `X.Y.Z` input. A bare git tag push does not
run the workflow. The build always checks out `vX.Y.Z`, not `main` HEAD.
`prepare` takes the version from the `workflow_dispatch` `version` input. A
dispatch whose input is missing or not `X.Y.Z` fails in `prepare`.
`prepare` takes the version from `github.event.release.tag_name` on a release
event, or from the `workflow_dispatch` `version` input. A dispatch whose input
is missing or not `X.Y.Z` fails in `prepare`.
```
release-on-publish (release: published → workflow_dispatch)
prepare (version from workflow_dispatch input; checkout vX.Y.Z)
release: published
prepare (version from release tag or workflow_dispatch input; checkout vX.Y.Z)
→ build [matrix: ubuntu-latest / ubuntu-24.04-arm]
→ docker build --load (VERSION from prepare)
→ e2e (test/e2e)
→ push ghcr.io/...:X.Y.Z-amd64 | X.Y.Z-arm64
→ merge
→ docker buildx imagetools create → unified manifest X.Y.Z
→ imagetools rm → drop X.Y.Z-amd64 and X.Y.Z-arm64 from GHCR
```
Native per-arch matrix (no QEMU): running the full Postfix/OpenDKIM stack under
emulation for e2e is impractical. E2e first, then push — the registry receives
the bytes that passed the gate.
the bytes that passed the gate. Only `ghcr.io/mixeme/selfpost:X.Y.Z` remains
tagged in GHCR; per-arch names exist briefly during the merge job.
A failed e2e **blocks** image publication.