feat(images): mirror the kagent line, the Substrate line and the Flux controllers under their upstream paths - #1235
Merged
Conversation
… controllers under their upstream paths
The agent-platform charts pull three sets of images from ghcr.io that have
no gsoci copy. Mirror them with the destination keeping the upstream path,
so a consumer changes one registry value per line and keeps the
repositories and digests the charts publish:
- the Giant Swarm kagent line, ghcr.io/giantswarm/kagent/{controller,ui,
golang-adk,claude-harness} -> gsoci.azurecr.io/giantswarm/kagent/<image>,
release tags X.Y.Z-gs.N from 0.11.0-gs.16 (images/renamed-kagent.yaml);
- the Giant Swarm Substrate line, ghcr.io/giantswarm/substrate/{ateapi,
atecontroller,atelet,atenet,podcertcontroller,ateom-gvisor} ->
gsoci.azurecr.io/giantswarm/substrate/<component>, release tags from
0.0.30-gs.4 (new images/renamed-substrate.yaml, added to the matrix);
- the Flux controllers ghcr.io/fluxcd/{source,helm}-controller ->
gsoci.azurecr.io/giantswarm/fluxcd/<controller> from the Flux 2.9 line
(source-controller v1.9, helm-controller v1.6), next to the flattened
fluxcd-* copies the flux-operator cannot compose from one registry value.
The release-tag filter with a prerelease floor is the agentgateway line's
pattern; dev builds and SBOM tags stay in ghcr. A slash in
override_repo_name already keeps a nested path (the strimzi entries), so
no code change: the field's comment now says so.
Tests: TestFilterTags pins the tag selection of these shapes;
TestRenamedImagesFiles parses and validates every images/renamed-*.yaml
the way `retagger run` does, so a broken entry fails in CI.
teemow
deleted the
feat/mirror-kagent-substrate-fluxcd-upstream-paths
branch
September 18, 2026 22:05
This file contains hidden or bidirectional Unicode text that may be interpreted or compiled differently than what appears below. To review, open the file in an editor that reveals hidden Unicode characters.
Learn more about bidirectional Unicode characters
Sign up for free
to join this conversation on GitHub.
Already have an account?
Sign in to comment
Add this suggestion to a batch that can be applied as a single commit.This suggestion is invalid because no changes were made to the code.Suggestions cannot be applied while the pull request is closed.Suggestions cannot be applied while viewing a subset of changes.Only one suggestion per line can be applied in a batch.Add this suggestion to a batch that can be applied as a single commit.Applying suggestions on deleted lines is not supported.You must change the existing code in this line in order to create a valid suggestion.Outdated suggestions cannot be applied.This suggestion has been applied or marked resolved.Suggestions cannot be applied from pending reviews.Suggestions cannot be applied on multi-line comments.Suggestions cannot be applied while the pull request is queued to merge.Suggestion cannot be applied right now. Please check back later.
Closes #1229.
What
Three sets of images the agent-platform charts pull from ghcr.io get a gsoci copy whose destination keeps the upstream path (nested repositories, as the strimzi entries already do), so a consumer changes one registry value per line and keeps the repositories and digests the charts publish:
ghcr.io/giantswarm/kagent/{controller,ui,golang-adk,claude-harness}gsoci.azurecr.io/giantswarm/kagent/<image>X.Y.Z-gs.Nfrom0.11.0-gs.16ghcr.io/giantswarm/substrate/{ateapi,atecontroller,atelet,atenet,podcertcontroller,ateom-gvisor}gsoci.azurecr.io/giantswarm/substrate/<component>X.Y.Z-gs.Nfrom0.0.30-gs.4ghcr.io/fluxcd/{source-controller,helm-controller}gsoci.azurecr.io/giantswarm/fluxcd/<controller>>= v1.9.0/>= v1.6.0(the Flux 2.9 line)images/renamed-kagent.yaml; Substrate in a newimages/renamed-substrate.yaml(added to theretag-renamed-imagesmatrix); the Flux controllers inimages/renamed-images.yamlnext to the flattenedfluxcd-*copies, which the flux-operator cannot compose from one registry value.filterwith a prereleasesemverfloor is the agentgateway line's pattern: dev builds (X.Y.Z-dev.…) and SBOM tags (sha256-….sbom) stay in ghcr.golang-adkandclaude-harnessby the index digest stamped at publish.skopeo copy --allcopies the index unchanged, so the digest resolves at the destination as well.override_repo_nameis used verbatim in the destination, so a slash keeps a nested path. Its comment (inmain.goand the README copy) now says so.Tests
TestFilterTagspins the tag selection of these shapes: the-gs.Nfilter with a prerelease floor keeps0.11.0-gs.16,0.11.0-gs.20,0.11.1-gs.1, drops0.11.0-gs.15, dev and SBOM tags; the plain>= v1.9.0floor drops release candidates and non-semver tags.TestRenamedImagesFilesparses everyimages/renamed-*.yamlthe wayretagger rundoes and validates each entry (semver constraint, filter and pattern compile), so a broken entry fails in CI rather than in the nightly run.go test ./...,golangci-lint(pre-commit) andyamllint(thevalidate-images-yamlconfig) pass locally.Verification after merge
build_retagonmaincopies the tags. Thencrane digeston source and destination must agree forgolang-adkandclaude-harnessat the chart's stamped digests, andcrane manifestmust resolve every Substrate component at0.0.30-gs.4and the two Flux controllers. The evidence goes on #1229.