Helm charts does not take into account the frontend / backend split

[MozeBaltyk]

currently I add to use the extraDeploy to deploy frontend apart from backend... it would be worth to add a template in the charts to handle this use case. I guess I am not the only to wonder how to deploy Frontend/backend backstage with an helm charts ?

Second point (more related to source code), the helm charts allow to create a configmaps "backstage-app-config" which I tried to reuse in the frontend but seems not be imported by frontend package/app :

          containers:
            - name: frontend
              image: localhost:5000/backstage-frontend:local
              imagePullPolicy: Always
              # Trying to use the same config maps created by backend - but not working
              envFrom:
                - configMapRef:
                    name: backstage-app-config

thrid point, would be better integration for ingress ...

Here is , how currently I deploy :

backstage:
  image:
    registry: localhost:5000
    repository: backstage-backend
    tag: local
    pullPolicy: Always

  startupProbe:
    failureThreshold: 30
    initialDelaySeconds: 60
    periodSeconds: 10
    successThreshold: 1
    timeoutSeconds: 2

  # Allow to load app-config.<BACKSTAGE_ENV>.yaml.
  extraEnvVars:
    - name: BACKSTAGE_ENV
      value: k3d

  # Create an backstage-app-config config-maps
  appConfig:
    app:
      title: Backstage
      baseUrl: http://backstage-frontend.backstage.svc.cluster.local
    backend:
      baseUrl: http://backstage.backstage.svc.cluster.local
      cors: 
        origin: http://backstage-frontend.backstage.svc.cluster.local

  # kubectl create secret generic backstage-secrets --from-env-file=vanilla/.env -n backstage
  # extraEnvVarsSecrets: 
  #   - backstage-secrets

  podSecurityContext:
    runAsUser: 65532
    runAsNonRoot: true
    fsGroup: 65532

  containerSecurityContext:
    allowPrivilegeEscalation: false
    readOnlyRootFilesystem: true
    capabilities:
      drop:
        - ALL

  extraVolumes:
    - name: tmp
      emptyDir: {}

  extraVolumeMounts:
    - name: tmp
      mountPath: /tmp

# Postgre Block 
postgresql:
  # -- Switch to enable or disable the PostgreSQL helm chart
  enabled: true
  # -- The authentication details of the Postgres database
  auth:
    username: backstage
    password: backstagepassword
    postgresPassword: adminpassword

extraDeploy:
  # -----------------------------
  # Frontend Deployment
  # -----------------------------
  - apiVersion: apps/v1
    kind: Deployment
    metadata:
      name: backstage-frontend
      labels:
        app: backstage-frontend
    spec:
      replicas: 1
      selector:
        matchLabels:
          app: backstage-frontend
      template:
        metadata:
          labels:
            app: backstage-frontend
        spec:
          securityContext:
            runAsUser: 65532
            runAsNonRoot: true
          containers:
            - name: frontend
              image: localhost:5000/backstage-frontend:local
              imagePullPolicy: Always
              # env:
              #   - name: APP_CONFIG_app_title
              #     value: Backstage
              #   - name: APP_CONFIG_app_baseUrl
              #     value: https://backstage.io
              #   - name: APP_CONFIG_backend_baseUrl
              #     value: https://backstage.io

              # Trying to use the same config maps created by backend - but not working
              envFrom:
                - configMapRef:
                    name: backstage-app-config
              
              securityContext:
                allowPrivilegeEscalation: false
                capabilities:
                  drop:
                    - ALL
              ports:
                - name: http
                  containerPort: 8080
              # env:
              #   - name: PORT
              #     value: "3000"
              resources:
                requests:
                  cpu: 100m
                  memory: 128Mi
                limits:
                  cpu: 500m
                  memory: 256Mi
              readinessProbe:
                httpGet:
                  path: /
                  port: 8080
                initialDelaySeconds: 5
                periodSeconds: 10
              livenessProbe:
                httpGet:
                  path: /
                  port: 8080
                initialDelaySeconds: 15
                periodSeconds: 20

  # -----------------------------
  # Frontend Ingress
  # -----------------------------
  - apiVersion: networking.k8s.io/v1
    kind: Ingress
    metadata:
      name: backstage-ingress
    spec:
      ingressClassName: traefik
      rules:
        - host: backstage.localhost
          http:
            paths:
              - path: /api
                pathType: Prefix
                backend:
                  service:
                    name: release-name-backstage
                    port:
                      number: 7007

              - path: /
                pathType: Prefix
                backend:
                  service:
                    name: backstage-frontend
                    port:
                      number: 3000

  # -----------------------------
  # Frontend Service
  # -----------------------------
  - apiVersion: v1
    kind: Service
    metadata:
      name: backstage-frontend
    spec:
      type: ClusterIP
      selector:
        app: backstage-frontend
      ports:
        - name: http-frontend
          port: 3000
          targetPort: 8080

[younsl]

I have the same need. Currently considering the Delivery Hero Backstage chart which has separate frontend/backend deployments. Would be nice if the official chart supported this natively.

[sid200727]

Hi,

Thanks for raising this the frontend/backend split makes a lot of sense for real deployments.

From what I understand, the current Helm chart assumes a single deployment that bundles both frontend and backend, which makes it harder to run them independently.

One possible approach could be to introduce configurable values such as:
• frontend.enabled
• backend.enabled

and generate separate deployments/services when both are enabled, while still keeping the current behaviour as the default for simpler setups.

I’d be happy to explore a PR for this if the maintainers think this direction aligns with the intended chart structure.

Would love to hear your thoughts before starting implementation.