Overture Helm-charts

Repository for Helm charts used by the Overture stack.

Using these charts

Charts are published as OCI artifacts to ghcr.io/overture-stack/helm-charts. No helm repo add step is required.

Helm CLI:

helm install <release-name> oci://ghcr.io/overture-stack/helm-charts/<chart-name> --version <version>

Terraform (hashicorp/helm provider):

resource "helm_release" "example" {
  repository = "oci://ghcr.io/overture-stack/helm-charts"
  chart      = "<chart-name>"
  version    = "<version>"
  namespace  = "<namespace>"
  # ...
}

Publishing

Bump version in <chart-name>/Chart.yaml and push to main. Jenkins lints, packages, and publishes automatically. See DEVELOPMENT.md for the full workflow and local testing commands.

---

Upgrade notes

Arranger

Version 0.4.1

New value: config.elasticsearch.password.secretKey - explicit key name to read from the password secret. When omitted, the chart falls back to using the username value as the key (backwards-compatible with 0.4.0).

Pods now automatically restart when catalogue configs change. A checksum/config annotation on the pod template is computed from config.catalogues at render time; Helm triggers a rolling restart whenever it changes.

Version 0.4.0 — breaking changes

This release aligns the chart with the current Arranger search-server and introduces multi-catalogue support.

Multi-catalogue config (replaces config.components.*):

Config files are now per-catalogue. Replace config.components.* with a config.catalogues list:

# Old (pre-0.4.0)
config:
  components:
    baseConfigs: { ... }
    tableConfigs: { ... }

# New
config:
  catalogues:
    - id: my-catalogue   # DNS label: lowercase alphanumeric and hyphens only
      configs:
        base: { ... }
        table: { ... }

A single-catalogue deployment is one entry in the list. Each catalogue's config files are mounted at <config.path>/<id>/ and discovered automatically by the server.

Deploying with Terraform:

Pass config files as decoded JSON objects — the chart's toPrettyJson step requires objects, not raw strings.

values = [
  file("helm/arranger-my-catalogue/values.yaml"),
  jsonencode({
    config = {
      catalogues = [
        {
          id = "my-catalogue"
          configs = {
            extended = jsondecode(file("./helm/arranger-my-catalogue/configs/extended.json"))
            facets   = jsondecode(file("./helm/arranger-my-catalogue/configs/facets.json"))
            table    = jsondecode(file("./helm/arranger-my-catalogue/configs/table.json"))
          }
        }
      ]
    }
  })
]

The id becomes the ConfigMap name suffix (<release>-server-configs-<id>) and must be a valid DNS label. For multiple catalogues, add additional entries to the catalogues list with distinct id values.

Migrating from pre-0.4.0 Terraform:

The old pattern passed raw string file contents under top-level keys:

# Old — no longer works in 0.4.0
jsonencode({
  "extendedConfigs" = file("./configs/extended.json")
  "facetsConfigs"   = file("./configs/facets.json")
  "tableConfigs"    = file("./configs/table.json")
})

Replace with the config.catalogues structure above, wrapping each file(...) call with jsondecode.

Removed values:

Removed	Replacement
config.documentType	Set documentType in the catalogue's base.json
config.elasticsearch.index	Set esIndex in the catalogue's base.json
config.components.*	config.catalogues[].configs.*
baseConfigs, extendedConfigs, facetsConfigs, matchboxConfigs, tableConfigs (top-level fallbacks)	config.catalogues[].configs.*
apiConfig.*, apiImage.*, apiIngress.*	config.*, image.*, ingress.*
uiConfig.*, uiImage.*, uiIngress.*	UI sidecar removed

Renamed values:

Old	New
config.port	config.serverPort
config.maxDownloadRows	config.downloadMaxRows

New values:

Value	Description
config.catalogues[]	List of catalogues; each requires a DNS-label id and optional configs.*
config.features.*	Feature flags (disableDownloads, disableFilters, disableGraphqlPlayground, disableSets, allowCustomDownloadMaxRows)
config.graphql.maxAliases / maxDepth	GraphQL security limits; leave unset for no limit
config.elasticsearch.setsIndex / setsType	Arranger sets index configuration
config.allowedCorsOrigins	List of allowed CORS origins; omit to allow all
config.pingMs	Health check ping interval in milliseconds
config.searchEngine	elasticsearch or opensearch; leave empty to auto-detect

config.path is now absolute (/app/configs) and is used as both the CONFIGS_PATH env var and the container mount path. Override if your image uses a different location.

Version 0.3.1

all api<values> are now just <values>:

Old:

apiConfig:
  ...

apiImage:
  ...

apiIngress:
  ...

New:

config:
  ...

image:
  ...

ingress:
  ...

also, new required config values: config: documentType: "" (i.e. graphql field; e.g. "file") index: "" (i.e. Elasticsearch index/alias to query)

Version: 0.3.0

Values for ingress config changed:

Old:

apiIngress:
  enabled: true
  host: host.local

uiIngress:
  enabled: true
  host: host.local

New:

apiIngress:
  enabled: true
  hosts:
    - host: host.local
      paths:
        - path: /
          pathType: ImplementationSpecific

uiIngress:
  enabled: true
  hosts:
    - host: host.local
      paths:
        - path: /
          pathType: ImplementationSpecific

DMS-ui

Version: 1.1.0

Old:

ingress:
  enabled: true
  hosts:
  - host.local

New:

ingress:
  enabled: true
  hosts:
    - host: host.local
      paths:
        - path: /
          pathType: ImplementationSpecific

Ego

Version: 3.1.0

Old:

ingressApi:
  enabled: true
  host: host.local

ingressUi:
  enabled: true
  host: ui.host.local

New:

ingressApi:
  enabled: true
  hosts:
    - host: ego.local
      paths:
        - path: /api/(.*)
          pathType: ImplementationSpecific

ingressUi:
  enabled: true
  hosts:
    - host: ui.ego.local
      paths:
        - path: /
          pathType: ImplementationSpecific

Lectern

Version: 0.7.0

Old:

ingress:
  enabled: true
  hosts:
  - host.local

New:

ingress:
  enabled: true
  hosts:
    - host: host.local
      paths:
        - path: /
          pathType: ImplementationSpecific

Maestro

See examples on how to use this chart in ./_examples

Version: 0.9.0

Old:

ingress:
  enabled: true
  hosts:
  - host.local

New:

ingress:
  enabled: true
  hosts:
    - host: host.local
      paths:
        - path: /
          pathType: ImplementationSpecific

Muse

Version: 0.10.0

Old:

ingress:
  enabled: true
  hosts:
  - host.local

New:

ingress:
  enabled: true
  hosts:
    - host: host.local
      paths:
        - path: /
          pathType: ImplementationSpecific

Rollcall

Version: 1.5.0

Old:

ingress:
  enabled: true
  hosts:
  - host.local

New:

ingress:
  enabled: true
  hosts:
    - host: host.local
      paths:
        - path: /
          pathType: ImplementationSpecific

Score

Version: 0.12.0

Old:

ingress:
  enabled: true
  hosts:
  - host.local

New:

ingress:
  enabled: true
  hosts:
    - host: host.local
      paths:
        - path: /
          pathType: ImplementationSpecific

Song

Version: 0.12.3

Allows using your own existing postgres instance. Set useInternalDep as false to prevent starting a "song-postgres" pod.

New params: (and their default values)

postgres:
  host: "song.fullname"-postgres.
  port: 5432
  secret:
    name: "song.fullname"-postgres
    key: postgres-password
  useInternalDep: true

Version: 0.12.0

Old:

ingress:
  enabled: true
  hosts:
  - host.local

New:

ingress:
  enabled: true
  hosts:
    - host: host.local
      paths:
        - path: /
          pathType: ImplementationSpecific

Stateless-svc

Version: 0.1.0

Old:

ingress:
  enabled: true
  hosts:
  - host.local

New:

ingress:
  enabled: true
  hosts:
    - host: host.local
      paths:
        - path: /
          pathType: ImplementationSpecific