Developer guide

Data pipeline

How CRDs, OpenAPI specs, dependency graphs, release notes and the sitemap are produced and when each step runs.

All the data the site shows is produced ahead of time by scripts and committed under static/. Adding an EDA release means running this pipeline against a cluster that has that release installed.

End to end

flowchart TB
  subgraph Cluster["EDA cluster with the release installed"]
    CRD["kubectl get crds *.eda.nokia.com"]
    MAN["EDA manifests in eda-system"]
    OAPI["EDA API /openapi/v3"]
  end

  CRD --> FETCH["fetch-crds-for-release.sh"]
  MAN --> FETCH
  FETCH --> YAML["static/resources/{release}/{crd}/{version}.yaml"]
  YAML --> BACK["Backfill from sibling patch releases"]
  BACK --> GEN["generate-manifests.sh"]
  GEN --> MJ["manifest.json + releases.yaml"]
  MJ --> DG["build-dependency-graph.ts"]
  DG --> DGJ["dependency-graph.json"]
  MJ --> RN["generate-release-notes.ts"]
  RN --> RNJ["static/release-notes/*.json"]

  OAPI --> SYNC["sync-openapi.sh"]
  SYNC --> SPECS["static/openapi/{release}/..."]
  SPECS --> OM["generate-openapi-manifest.ts"]
  OM --> OMJ["openapi manifest.json + openapi-releases.yaml"]
  SPECS --> IDX["build-openapi-index.ts"]
  IDX --> IDXJ["api-index.json"]

  MJ --> SM["generate-sitemap.mjs"]
  SM --> SMX["sitemap.xml, robots.txt"]

Steps

Step	Command	Output	When
Sync a CRD release	`static/sync-release.sh {release}`	Runs the next four steps in order	Manual, per new release
Fetch CRDs	`static/lib/fetch-crds-for-release.sh`	One YAML file per CRD version	Via sync-release
Backfill	`static/lib/backfill-crds-lib.sh`	Copies CRDs missing from a patch release from its siblings, with a `.backfilled-from` marker	Via sync-release
Manifests	`static/generate-manifests.sh`	`manifest.json` per release; updates `src/lib/releases.yaml`	Via sync-release or manual
Dependency graph	`npm run build:dependency-graph`	`dependency-graph.json` per release	Via sync-release or manual
Release notes	`npm run generate:release-notes`	`static/release-notes/{release}.json`, `index.json`, `bundle.json`	`prebuild` and sync-release
OpenAPI sync	`RELEASE={release} npm run sync:openapi`	Raw discovery plus `core/` and `apps/{group}/{version}/` specs	Manual
OpenAPI manifest	`npm run generate:openapi-manifest -- --all`	`static/openapi/{release}/manifest.json`; updates `src/lib/openapi-releases.yaml`	`prebuild`
API index	`npm run build:openapi-index`	`static/openapi/{release}/api-index.json`: each API's resources, workflows and sections	`prebuild`
Sitemap	`npm run generate:sitemap`	`static/sitemap.xml`, `static/robots.txt`	`prebuild`
CRD scope	`npm run build:crd-scopes`	`scope` (Namespaced / Cluster) in each `manifest.json`, from the OpenAPI paths	`generate:all`
Dependency audit	`RELEASE={release} npm run audit:dependencies`	`dependency-audit.json` / `.md` report	Manual

sync-release.sh accepts --dry-run, --backfill-only and --skip-* flags to run parts of the chain.

Warning sync:openapi needs RELEASE set. Without it, specs land in static/openapi/unlabeled/, which the Cloudflare build deletes as a safety measure.

Data formats

manifest.json — the index of a release:

[
	{
		"name": "interfaces.interfaces.eda.nokia.com",
		"group": "interfaces.eda.nokia.com",
		"kind": "Interface",
		"versions": [{ "name": "v1" }, { "name": "v1alpha1", "deprecated": true }]
	}
]

CRD YAML — one file per CRD version, containing that version's entry from spec.versions[], including schema.openAPIV3Schema.

releases.yaml — ordered release list; one entry has default: true:

releases:
  - name: '26.8.2'
    label: 'EDA 26.8.2'
    folder: 'resources/26.8.2'
    default: true

Sitemap contents

generate-sitemap.mjs writes the tool pages, every CRD of the default release (as /?release=…&crd=…&version=…, matching each page's canonical URL), and every docs page. Kinds that exist in several API groups use the full CRD name so each URL opens exactly one CRD.

Data pipeline

#End to end

#Steps

#Data formats

#Sitemap contents

End to end

Steps

Data formats

Sitemap contents