A component in AICR is a registry entry pointing to a Helm chart or
Kustomize source that recipes can pull. The catalog lives in
recipes/registry.yaml;
per-component default values live under
recipes/components/<name>/.
Overlays bind a component to a cluster shape; bundlers turn that
binding into a deployer-specific artifact.
Most components need no Go code. The declarative path is one
registry entry plus a values.yaml. The legacy
pkg/component/generic.go::ComponentConfig
is marked Deprecated — it is unused in production. The live schema
is pkg/recipe/components.go::ComponentConfig.
For the recipe data model — overlays, mixins, criteria, merge order — see recipe.md. This page is the contributor view for adding or changing components.
| I want to... | Edit | Guide |
|---|---|---|
| Make an existing chart or kustomization available to recipes | recipes/registry.yaml entry |
this page |
| Set default values for the chart | recipes/components/<name>/values.yaml |
this page |
| Pin a chart version for a specific cluster shape | Recipe overlay in recipes/overlays/ |
recipe.md |
| Add a bundle-time validation warning | registry.yaml validations: block |
validator.md |
| Add a chainsaw health check | registry.yaml healthCheck.assertFile + recipes/checks/<name>/health-check.yaml |
validator.md |
| Adjust where node selectors land in chart values | registry.yaml nodeScheduling paths |
this page |
A component declares either helm: or kustomize: — never
both. ComponentRegistry.Validate rejects the mixed shape at load.
Use helm: when |
Use kustomize: when |
|---|---|
| Upstream ships a published Helm chart | Upstream ships only a Git source with kustomization.yaml |
You need --set value overrides |
You can accept no --set support |
You want nodeScheduling injection |
You will configure scheduling out-of-band (Kustomize ignores Helm value paths) |
Kustomize limitations to know up front:
--set <key>:<path>=<value>flows through Helm value rendering only; Kustomize components silently ignore overrides.nodeScheduling.system/acceleratedpaths target Helm values; they do not apply to Kustomize sources.- The bundler runs
kustomize buildat bundle time and wraps the output astemplates/manifest.yamlinside the standard local-format folder (see index.md for the classification rule).
1. Add the registry entry to recipes/registry.yaml:
- name: my-operator
displayName: My Operator
valueOverrideKeys:
- myoperator
helm:
defaultRepository: https://charts.example.com
defaultChart: example/my-operator
defaultVersion: v1.0.0
defaultNamespace: my-operator
nodeScheduling:
system:
nodeSelectorPaths: [operator.nodeSelector]
tolerationPaths: [operator.tolerations]2. Create recipes/components/my-operator/values.yaml with the
chart defaults you want every recipe to start from. Keep this file
minimal and widely applicable — cluster-specific tweaks belong in
values-<context>.yaml siblings referenced from an overlay.
# fullnameOverride avoids the aicr-stack- prefix on resource names.
fullnameOverride: my-operator
operator:
replicas: 13. Optional blocks on the registry entry:
validations:— bundle-time misconfiguration warnings (validator.md)healthCheck.assertFile:— chainsaw conformance assertions (validator.md)storageClassPaths:— where--storage-classis injectedpodScheduling.workload.workloadSelectorPaths— for workload-pod placementgkeCriticalPriority,hasSelfRefCRDs— narrow service-specific quirks (see godoc onComponentConfigfor when these apply)
4. Run make bom-docs and commit the regenerated
docs/user/container-images.md in the same PR. CLAUDE.md treats this
as a hard rule whenever you change registry.yaml, a component's
values.yaml, or any chart version pin. See BOM regeneration.
5. Run make qualify — covers tests, lint, and the recipe-resolution
suite that parses every registry entry.
- name: my-kustomize-app
displayName: My Kustomize App
valueOverrideKeys:
- mykustomize
kustomize:
defaultSource: https://github.com/example/my-app
defaultPath: deploy/production
defaultTag: v1.0.0No recipes/components/<name>/values.yaml is required — Kustomize
reads its inputs from the upstream source. Reminder: no --set
overrides, and nodeScheduling paths do not apply.
Authoritative definitions live in
pkg/recipe/components.go.
One-liner per field:
| Field | Purpose |
|---|---|
name |
Component identifier; must match componentRefs[].name in overlays |
displayName |
Human-readable label used in CLI output and bundle templates |
valueOverrideKeys |
Alt prefixes for --set <key>:path=value matching |
helm.defaultRepository |
Helm repo URL injected when an overlay leaves it empty |
helm.defaultChart |
Chart name (e.g. nvidia/gpu-operator) |
helm.defaultVersion |
Default chart version |
helm.defaultNamespace |
Install namespace |
kustomize.defaultSource |
Git or OCI source URL |
kustomize.defaultPath |
Subpath within the source |
kustomize.defaultTag |
Git ref / OCI tag |
nodeScheduling.system |
Helm value paths that receive the control-plane node selector / tolerations / taints |
nodeScheduling.accelerated |
Helm value paths that receive the GPU node selector / tolerations / taints |
nodeScheduling.nodeCountPaths |
Where --nodes is written |
podScheduling.workload.workloadSelectorPaths |
Workload-pod placement |
storageClassPaths |
Where --storage-class is written |
validations |
Bundle-time component check list (validator.md) |
healthCheck.assertFile |
Chainsaw assert YAML path (relative to data dir) |
gkeCriticalPriority, hasSelfRefCRDs |
Narrow service-specific flags (see godoc) |
This is the field most contributors get wrong on first PR.
system— paths into chart values for workloads that must land on management / control-plane nodes (e.g., operators, controllers, webhooks). The bundler writes the--system-node-selectorand--system-node-tolerationvalues here.accelerated— paths into chart values for workloads that must land on GPU nodes (e.g., device-plugin DaemonSets, driver-validation, DCGM exporters). The bundler writes the--accelerated-node-selectorand--accelerated-node-tolerationvalues here.
Concrete example from gpu-operator:
nodeScheduling:
system:
nodeSelectorPaths:
- operator.nodeSelector
- node-feature-discovery.master.nodeSelector
tolerationPaths:
- operator.tolerations
accelerated:
nodeSelectorPaths:
- daemonsets.nodeSelector
- node-feature-discovery.worker.nodeSelector
tolerationPaths:
- daemonsets.tolerationsWrong column = workloads land on the wrong node class. A DaemonSet
placed under system will miss GPU nodes; an operator under
accelerated will refuse to schedule on a cluster with tainted GPU
nodes only.
--set <key>:<path>=<value> matches via GetByOverrideKey:
- The component
nameis checked first. - Each entry in
valueOverrideKeysis then checked.
For gpu-operator with valueOverrideKeys: [gpuoperator], both
--set gpu-operator:driver.version=... and
--set gpuoperator:driver.version=... resolve to the same component.
Pick a key that is easier to type (no hyphens) and document it in the
displayName-adjacent comments if non-obvious. Override keys are
globally unique — ComponentRegistry.Validate rejects duplicates.
A recurring first-PR instinct is to add a new CLI flag for every placement knob a chart exposes — one flag for the controller, one for the workers, one for a sidecar, and so on. Don't. AICR already models node placement as value paths, not flags, so the existing primitives cover arbitrary chart depth without growing the CLI surface.
Two separate mechanisms are at work — keep them distinct:
Path routing. nodeScheduling is registry metadata that maps chart
value paths onto the system / accelerated node classes (see
the section above). It selects
which chart paths receive a node selector / toleration; it does not
participate in value precedence. The bundler fans --system-node-selector
/ --accelerated-node-selector (and the matching *-toleration flags)
out to every declared path, so one flag covers N workloads.
Value precedence. Once a path is targeted, its value is resolved across layers, lowest precedence first:
- Component value defaults — ship sane defaults in
recipes/components/<name>/values.yaml, or per-recipe in an overlaycomponentRefs[].valuesFile/ inlineoverrides. This is where the common case (e.g. a managed cluster's standard node labels) works out of the box with no flags at all. - Deploy-time overrides —
--set <key>:<path>=<value>reaches any value path for last-mile deviations;--dynamic <key>:<path>defers a path to install time (it is stripped fromvalues.yamlintocluster-values.yamlfor the operator to fill in).
Merge order is base → valuesFile → overlay overrides →
--set / --dynamic, so a bundle-time override always wins over a
recipe default. The node-class flags (--system-node-selector et al.)
write into their routed paths at bundle time, alongside --set.
Deciding where a knob belongs:
| Situation | Where it goes | Why |
|---|---|---|
| Workload must run on management / control-plane nodes | Path under nodeScheduling.system |
Inherits --system-node-selector automatically |
| Workload must run on GPU nodes | Path under nodeScheduling.accelerated |
Inherits --accelerated-node-selector automatically |
| A pool that is neither (e.g. a dedicated CPU-worker pool) | Leave it out of nodeScheduling; target via --set <key>:that.path.nodeSelector.<label>=<value> |
The two-class model intentionally covers only system vs GPU; everything else is reachable by path |
| Value is cluster-specific and unknown at bundle time | --dynamic <key>:<path> |
Lands in cluster-values.yaml for install-time entry |
"When omitted, inherit from the node class" needs no special logic — a
path listed under system / accelerated is the inheritance.
Caveat: the chart must actually render the path. A nodeSelector
path only takes effect if the upstream chart template renders it. Some
charts honor only affinity / tolerations and silently drop
nodeSelector, so a declared path becomes a no-op. Verify against the
chart's templates before adding a path. If the chart does not support
it, either pin placement through the affinity path the chart does
render (as a values.yaml default) or omit the path and document the
limitation in a registry comment. The slinky-slurm-operator entry in
registry.yaml
is a worked example of the latter.
RecipeResult.DeploymentOrder is derived, not authored.
TopologicalSort in pkg/recipe/metadata_store.go orders components
by componentRefs[].dependencyRefs declared in the overlay. When no
dependencies are declared, the order falls back to the order in
which components are listed in the overlay's componentRefs. Express
ordering by declaring dependencyRefs on the dependent component, not
by writing a separate deploymentOrder block.
The bundler emits a uniform NNN-<component>/ layout via
pkg/bundler/deployer/localformat.
Classification (single source of truth in localformat.classify):
| Recipe shape | Folder kind |
|---|---|
helm.defaultRepository set, no manifestFiles |
KindUpstreamHelm |
helm.defaultRepository set + manifestFiles |
KindUpstreamHelm + KindLocalHelm (-post injected) |
helm.defaultRepository == "" + manifestFiles |
KindLocalHelm |
kustomize.*Tag or *Path set |
KindLocalHelm (kustomize build → templates/manifest.yaml) |
If both helm and kustomize fields are populated, Validate
rejects the registry entry — there is no precedence rule because the
shape is invalid. manifestFiles are added post-chart; preManifestFiles
ship at sync-wave N-1 (e.g., a Namespace with PSS labels the chart
pods depend on).
AICR ships five output adapters in
pkg/bundler/deployer/:
helm, helmfile, argocd, argocdhelm, flux. Each calls
localformat.Write() and then layers its own orchestration files
(deploy.sh, helmfile.yaml, Argo Application CRs, Flux
HelmReleases). Components do not need to be deployer-aware —
the bundler renders per-deployer from one component definition.
See index.md for the deployer matrix.
docs/user/container-images.md is rendered fresh from each Helm
chart's actual templates by make bom-docs. Run it and commit the
regenerated file in the same PR whenever you:
- Add or remove a component
- Bump a chart version (in
registry.yaml, an overlay, or a mixin) - Change a
values.yamlin a way that affects which images render (image-repo override, subchart enable/disable, etc.)
make bom-check verifies the committed BOM matches a fresh regen
but is opt-in only — not wired into make qualify, make lint,
or the merge gate. Do not rely on CI to catch a missed regen.
A component entry describes what to deploy and where its values
land. Components do not carry apply, wait, uninstall, rollback,
or readiness-polling logic — those concerns belong to the deployer
that consumes the bundle. If you find yourself writing custom apply
code inside the bundler or under pkg/component/, you are on the
wrong side of the boundary — see
index.md "What AICR Is Not".
- recipe.md — overlays, mixins, criteria, the recipe data model
- validator.md — bundle-time component validation checks
- validator.md — chainsaw health checks and validator runner
- index.md — contributor index and architectural boundary
- integrator/recipe-development.md — end-user recipe authoring
- user/component-catalog.md — end-user component catalog
pkg/recipe/components.go—ComponentConfigsource of truthrecipes/registry.yaml— live component catalog