generate-manifests
Whenever you touch the API (api/v1, api/v1alpha1) or RBAC/kubebuilder markers in
controllers/, several files are generated from them and must be regenerated and
committed together with the source change. This skill covers the whole generation
chain, from cheap codegen to the full OLM bundle.
When to use
- Added/changed a field, type, or validation in
api/v1*/…_types.go.
- Changed a
// +kubebuilder:... marker (CRD schema, printer columns, RBAC).
- Edited a sample CR in
config/samples/ (feeds the CSV alm-examples).
- Need refreshed CRDs, RBAC role, CSV/bundle, or ArgoCD manifests before a PR.
Commands (cheapest first)
1. Code + CRD/RBAC generation
make generate # DeepCopy methods -> api/*/zz_generated.deepcopy.go
make manifests # CRDs -> config/crd/bases, RBAC role, CSV base annotations
Or both at once:
make manifests generate
2. Full OLM bundle (do this before a PR that changes the API)
make bundle
make bundle runs generate + manifests, then via pre-bundle:
operator-sdk generate kustomize manifests + kustomize build config/manifestsoperator-sdk generate bundle (writes bundle/manifests, bundle/metadata,
bundle/tests)- reorders owned CRDs so IBMLicensing is first, sets package/OpenShift-version
annotations, injects
relatedImages from common/relatedImages.yaml
- folds RBAC rules and
alm-examples from config/samples into the CSV
- validates the bundle with
operator-sdk bundle validate ./bundle
3. ArgoCD / GitOps YAMLs (only when the deployment shape changed)
make generate-yaml-argo-cd
Rebuilds the templated ArgoCD manifests (cluster-rbac, cr, crd, deployment, rbac,
serviceaccounts) under the deploy/argo-cd component sources from the kustomize output.
Needed only when the operator Deployment, RBAC, or CRDs change - not for pure API field
tweaks that don't alter RBAC.
Files produced / updated
api/v1/zz_generated.deepcopy.go, api/v1alpha1/zz_generated.deepcopy.goconfig/crd/bases/operator.ibm.com_*.yamlconfig/rbac/role.yamlconfig/manifests/bases/ibm-licensing-operator.clusterserviceversion.yamlbundle/manifests/… (CSV + CRDs), bundle/metadata/annotations.yaml
The owned CRDs (what the CRDs cover)
| Kind |
API version |
Scope |
| IBMLicensing |
v1alpha1 |
cluster-scoped |
| IBMLicensingMetadata |
v1alpha1 |
namespaced |
| IBMLicensingDefinition |
v1 |
namespaced |
| IBMLicensingQuerySource |
v1 |
namespaced |
Prerequisites
controller-gen, kustomize, yq, and operator-sdk installed - see [[setup-tools]].
Notes
- Always commit generated files with the API change (CONTRIBUTING.md requires it).
CSV_VERSION (default 4.2.24) drives the bundle version; overriding it changes CSV
contents - don't bump it as a side effect of regeneration.- The
temp/ directory holds intermediate build files and is gitignored; don't commit it.
- Version bumps use
common/scripts/next_csv.sh <current> <new> <old>, not this skill.
Related skills
- [[operator-sdk-guide]] - what the markers and generators actually do.
- [[code-quality]] / [[unit-test]] - run after regenerating.
- [[build-and-deploy]] - deploy the regenerated CRDs to a cluster.
1---2name: generate-manifests3description: Regenerate all derived artifacts after changing the ibm-licensing-operator API types or config - DeepCopy code, CRDs, RBAC, the OLM bundle/CSV, and the ArgoCD GitOps YAMLs. Use after editing anything in api/ or controllers/ (new fields, kubebuilder markers, RBAC markers, samples) so generated files stay in sync and can be committed.4---56# generate-manifests78Whenever you touch the API (`api/v1`, `api/v1alpha1`) or RBAC/kubebuilder markers in9`controllers/`, several files are generated from them and **must be regenerated and10committed** together with the source change. This skill covers the whole generation11chain, from cheap codegen to the full OLM bundle.1213## When to use1415- Added/changed a field, type, or validation in `api/v1*/…_types.go`.16- Changed a `// +kubebuilder:...` marker (CRD schema, printer columns, RBAC).17- Edited a sample CR in `config/samples/` (feeds the CSV `alm-examples`).18- Need refreshed CRDs, RBAC role, CSV/bundle, or ArgoCD manifests before a PR.1920## Commands (cheapest first)2122### 1. Code + CRD/RBAC generation2324```bash25make generate # DeepCopy methods -> api/*/zz_generated.deepcopy.go26make manifests # CRDs -> config/crd/bases, RBAC role, CSV base annotations27```2829Or both at once:3031```bash32make manifests generate33```3435### 2. Full OLM bundle (do this before a PR that changes the API)3637```bash38make bundle39```4041`make bundle` runs `generate` + `manifests`, then via `pre-bundle`:42- `operator-sdk generate kustomize manifests` + `kustomize build config/manifests`43- `operator-sdk generate bundle` (writes `bundle/manifests`, `bundle/metadata`,44 `bundle/tests`)45- reorders owned CRDs so **IBMLicensing is first**, sets package/OpenShift-version46 annotations, injects `relatedImages` from `common/relatedImages.yaml`47- folds RBAC rules and `alm-examples` from `config/samples` into the CSV48- validates the bundle with `operator-sdk bundle validate ./bundle`4950### 3. ArgoCD / GitOps YAMLs (only when the deployment shape changed)5152```bash53make generate-yaml-argo-cd54```5556Rebuilds the templated ArgoCD manifests (cluster-rbac, cr, crd, deployment, rbac,57serviceaccounts) under the `deploy/argo-cd` component sources from the kustomize output.58Needed only when the operator Deployment, RBAC, or CRDs change - not for pure API field59tweaks that don't alter RBAC.6061## Files produced / updated6263- `api/v1/zz_generated.deepcopy.go`, `api/v1alpha1/zz_generated.deepcopy.go`64- `config/crd/bases/operator.ibm.com_*.yaml`65- `config/rbac/role.yaml`66- `config/manifests/bases/ibm-licensing-operator.clusterserviceversion.yaml`67- `bundle/manifests/…` (CSV + CRDs), `bundle/metadata/annotations.yaml`6869## The owned CRDs (what the CRDs cover)7071| Kind | API version | Scope |72|------|-------------|-------|73| IBMLicensing | v1alpha1 | cluster-scoped |74| IBMLicensingMetadata | v1alpha1 | namespaced |75| IBMLicensingDefinition | v1 | namespaced |76| IBMLicensingQuerySource | v1 | namespaced |7778## Prerequisites7980controller-gen, kustomize, yq, and operator-sdk installed - see [[setup-tools]].8182## Notes8384- **Always commit generated files** with the API change (CONTRIBUTING.md requires it).85- `CSV_VERSION` (default `4.2.24`) drives the bundle version; overriding it changes CSV86 contents - don't bump it as a side effect of regeneration.87- The `temp/` directory holds intermediate build files and is gitignored; don't commit it.88- Version bumps use `common/scripts/next_csv.sh <current> <new> <old>`, not this skill.8990## Related skills9192- [[operator-sdk-guide]] - what the markers and generators actually do.93- [[code-quality]] / [[unit-test]] - run after regenerating.94- [[build-and-deploy]] - deploy the regenerated CRDs to a cluster.
IBM