Forte/launchpad
6
0
Fork 0
Code Issues Pull Requests Actions Packages Wiki Activity
Files
21fb50ba00b4e3dcfa5580c0d7ba0e503fece89c
launchpad/docs/REFERENCE.md
Danijel Simeunovic 21fb50ba00 vw fixes
2026-05-16 15:55:18 +02:00

62 KiB
Raw Blame History

Technical Reference

Table of Contents

Architecture Components

Cluster Specifications

Component Value
Provider Multi-cloud (UpCloud, AWS EKS, Azure AKS, GCP GKE)
Environment Dev + Production per cloud
Active clusters UpCloud (upc-dev, upc-prod)
Cloud-ready templates EKS, AKS, GKE (dev + prod each)
GitOps Tool ArgoCD
Ingress Controller Traefik v2
Certificate Management Cert-Manager + Let's Encrypt
Policy Engine Kyverno
Secret Management Sealed Secrets (Bitnami)
Monitoring Prometheus + Grafana
Logging Loki + Fluent-Bit
Tracing Tempo (OTLP)
Container Scanning Trivy
Version Control Gitea

Network Architecture

Internet
   │
   ▼
[DNS: *.forteapps.net]
   │
   ▼
[Cloud Load Balancer]
   │
   ▼
[Traefik Ingress Controller]
   │
   ├──► IngressRoute (TLS termination via Cert-Manager)
   │
   ├──► Service (ClusterIP)
   │    │
   │    └──► Pod (Application Container)
   │
   └──► Service (Database - ClusterIP)
        │
        └──► StatefulSet (PostgreSQL)

Repository Reference

Config Repository: launchpad

URL: https://git.forteapps.net/Forte/launchpad

Directory Structure

launchpad/
├── bootstrap.sh                   # Cluster initialization script
├── _app-of-apps-upc-dev.yaml     # Root ArgoCD Application (upc-dev)
├── _app-of-apps-upc-prod.yaml    # Root ArgoCD Application (upc-prod)
│
├── infra/                         # Infrastructure applications (Kustomize)
│   ├── base/                      # One subdirectory per component
│   │   ├── kustomization.yaml     # Aggregates all component subdirectories
│   │   ├── traefik-application/
│   │   │   ├── kustomization.yaml
│   │   │   └── traefik-application.yaml
│   │   ├── keycloak/
│   │   │   ├── kustomization.yaml
│   │   │   └── keycloak.yaml
│   │   ├── grafana/
│   │   ├── prometheus/
│   │   ├── loki/
│   │   ├── tempo/
│   │   ├── gitea/
│   │   ├── opencost/
│   │   ├── ...                    # Each component in own directory
│   │   └── secrets/
│   ├── overlays/                  # Per-cluster: include all or cherry-pick
│   │   ├── upc-dev/              # resources: [../../base] (all components)
│   │   ├── upc-prod/            # resources: [../../base] + patches
│   │   ├── aks-dev/             # resources: [../../base/grafana, ...] (selective)
│   │   └── .../                  # 8 clusters total
│   └── values/
│       ├── base/                   # Cloud-agnostic Helm values
│       │   ├── gitea-values.yaml
│       │   ├── opencost-values.yaml
│       │   ├── prometheus-values.yaml
│       │   └── ...
│       ├── upc-dev/               # UpCloud dev overlay values
│       │   ├── traefik-values.yaml
│       │   ├── keycloak-values.yaml
│       │   ├── grafana-values.yaml
│       │   ├── gitea-values.yaml
│       │   └── opencost-values.yaml
│       └── upc-prod/              # UpCloud prod overlay values
│           ├── traefik-values.yaml
│           ├── keycloak-values.yaml
│           ├── grafana-values.yaml
│           ├── gitea-values.yaml
│           └── opencost-values.yaml
│
├── apps/                          # Business applications (Kustomize)
│   ├── base/                     # One subdirectory per app
│   │   ├── kustomization.yaml
│   │   ├── musicman/
│   │   ├── mcp10x/
│   │   ├── dot-ai-stack/
│   │   ├── ts-mcp/
│   │   └── argo-mcp/
│   └── overlays/                 # Per-cluster: include all or cherry-pick
│       ├── upc-dev/
│       ├── upc-prod/
│       └── aks-dev/              # Selective apps only
│
├── cluster-resources/             # Cluster-level resources
│   ├── cert-manager-namespace.yaml
│   ├── secrets-namespace.yaml
│   ├── letsencrypt-issuer.yaml
│   ├── kyverno-config.yaml
│   ├── argocd-notifications-secret-sealed.yaml
│   ├── forte10x-repo-credentials-sealed.yaml
│   ├── mcp10x-repo-credentials-sealed.yaml
│   └── policies/
│       ├── deployment-verifier.yaml
│       ├── label-checker.yaml
│       ├── bare-pod-cleaner.yaml
│       ├── replicaset-cleaner.yaml
│       ├── default-ns-blocker.yaml
│       ├── secret-cloner.yaml
│       ├── keycloak-client-cloner.yaml
│       └── auth-sidecar-injector.yaml
│
├── secrets/                       # Application secrets (sealed)
│   ├── base/                     # All SealedSecrets (shared across clouds)
│   │   ├── kustomization.yaml
│   │   ├── argocd-forte-helm-secret-sealed.yaml
│   │   ├── argocd-mcp-credentials.yaml
│   │   ├── argocdmcp-auth-oidc-sealed.yaml
│   │   ├── dot-ai-secrets.yaml
│   │   ├── forte10x-app-credentials-sealed.yaml
│   │   ├── gitea-backup-s3-sealed.yaml
│   │   ├── gitea-credentials-sealed.yaml
│   │   ├── gitea-runner-token-sealed.yaml
│   │   ├── gitea-smtp-secret-sealed.yaml
│   │   ├── keycloak-credentials-sealed.yaml
│   │   ├── musicman-auth-oidc-sealed.yaml
│   │   ├── musicman-credentials.yaml
│   │   └── renovate-env-sealed.yaml
│   └── overlays/                 # Per-cloud overlays (reference base)
│       ├── aks-dev/kustomization.yaml
│       ├── aks-prod/kustomization.yaml
│       ├── eks-dev/kustomization.yaml
│       ├── eks-prod/kustomization.yaml
│       ├── gke-dev/kustomization.yaml
│       ├── gke-prod/kustomization.yaml
│       ├── upc-dev/kustomization.yaml
│       └── upc-prod/kustomization.yaml
│
├── scripts/                       # Operational helper scripts
│   ├── gitea-backup.sh            # S3 backup helper (list/download)
│   ├── gitea-restore.sh
│   └── backup/                    # Per-cloud backup reference scripts
│       ├── s3-minio.sh            # S3-compatible (UpCloud, MinIO, Wasabi)
│       ├── aws-s3.sh              # Native AWS S3
│       ├── azure-blob.sh          # Azure Blob Storage
│       └── gcp-gcs.sh             # GCP Cloud Storage
│
├── private/                       # Local-only (Git-ignored)
│   ├── *.yaml
│   └── *.sh
│
└── docs/                          # Documentation
    ├── GITOPS-ARCHITECTURE.md
    ├── DEVELOPER-GUIDE.md
    ├── OPERATIONS-RUNBOOK.md
    └── REFERENCE.md

Key Files

bootstrap.sh

#!/bin/zsh
# Initializes cluster with ArgoCD

ArgoCd() {
  helm upgrade --install argocd argo-cd \
    --repo https://argoproj.github.io/argo-helm \
    --namespace argocd --create-namespace \
    --values infra/values/base/argocd-values.yaml \
    --set notifications.context.clusterName="$CLUSTER_NAME" \
    --timeout 60s --atomic

  kubectl apply -f _app-of-apps-upc-dev.yaml -n argocd   # or _app-of-apps-upc-prod.yaml
}

_app-of-apps-upc-dev.yaml / _app-of-apps-upc-prod.yaml

apiVersion: argoproj.io/v1alpha1
kind: Application
metadata:
  name: infrastructure-apps
  namespace: argocd
spec:
  project: default
  source:
    repoURL: ssh://git@git.forteapps.net:2222/Forte/launchpad.git
    path: infra
  destination:
    server: https://kubernetes.default.svc
    namespace: default
  syncPolicy:
    automated:
      prune: true
      selfHeal: true

Helm Charts Repository: forte-helm

URL: https://git.forteapps.net/Forte/forte-helm

Chart: forteapp

Version: 0.1.0 App Version: 1.0.0 Type: application

Templates
Template Purpose
_helpers.tpl Template helper functions
namespace.yaml Namespace resource
deployment.yaml Main application Deployment
service.yaml ClusterIP Service
ingressroute.yaml Traefik IngressRoute
certificate.yaml Cert-Manager Certificate
configmap.yaml Application ConfigMap
secret-auth-tokens.yaml Authentication tokens
hpa.yaml Horizontal Pod Autoscaler
database-statefulset.yaml Optional PostgreSQL StatefulSet
database-service.yaml PostgreSQL Service
Default Values Schema
app:
  image:
    repository: ""              # Required
    tag: ""                     # Required
    pullPolicy: IfNotPresent
    containerPort: 3000

  replicaCount: 1

  resources:
    requests:
      cpu: 100m
      memory: 128Mi
    limits:
      cpu: 500m
      memory: 512Mi

  hpa:
    enabled: false
    minReplicas: 2
    maxReplicas: 10
    targetCPUUtilizationPercentage: 70

  extraEnv: []
  # - name: KEY
  #   value: "value"

  envSecretName: ""             # Reference to Secret
  nodeEnv: production

db:
  enabled: false
  name: postgres
  image:
    repository: postgres
    tag: "16-alpine"

  service:
    type: ClusterIP
    port: 5432
    targetPort: 5432

  persistence:
    enabled: true
    storageClass: ""
    accessMode: ReadWriteOnce
    size: 5Gi

  resources:
    requests:
      memory: "256Mi"
      cpu: "250m"
    limits:
      memory: "1Gi"
      cpu: "1000m"

  extraEnv: []
  envSecretName: ""

  livenessProbe:
    exec:
      command:
      - pg_isready
      - -U
      - db_user
      - -d
      - db_name
    initialDelaySeconds: 30
    periodSeconds: 10

  readinessProbe:
    exec:
      command:
      - pg_isready
      - -U
      - db_user
      - -d
      - db_name
    initialDelaySeconds: 5
    periodSeconds: 5

service:
  type: ClusterIP
  port: 3000

ingress:
  enabled: false
  host: ""
  entrypoint: websecure
  tls:
    enabled: true
    secretName: ""
    clusterIssuer: letsencrypt-prod

auth:
  enabled: false                 # Enable authentication sidecar injection
  type: token                    # Authentication mode: "token" or "oidc"

  # Token-based authentication configuration
  tokens: []                     # List of valid bearer tokens (hex strings, 32+ bytes recommended)
  # - d4f88f6d9292c10cc3e21c4aad56d2be485db532b54fe961d738e1137d247823
  # - 8803f621acc3898df1d7a8f514bc3602551a0681a8f747bd4e43c3c5849d57a7

  # OIDC authentication configuration
  oidc:
    authority: ""                # OIDC provider URL (e.g., https://auth.example.com/realms/master)
    clientId: ""                 # OIDC client ID registered with provider
    scopes: "openid,profile,email"  # OAuth scopes (comma-separated)
    callbackPath: /auth/callback    # OAuth callback path (default: /auth/callback)
    # Note: Client secret must be in 'auth-oidc' Secret (client-secret key)
    #       Cookie secret must be in 'auth-oidc' Secret (cookie-secret key)

configmap: []                    # Application ConfigMap key-value pairs
# KEY: value
# DB_HOST: postgres
# DB_PORT: "5432"

Helm Values Repository: helm-prod-values

URL: https://git.forteapps.net/Forte/helm-prod-values.git

Structure

helm-prod-values/
├── mcp10x/
│   └── values.yaml
├── musicman/
│   └── values.yaml
└── argocd-mcp/
    └── values.yaml

Example: mcp10x/values.yaml

app:
  image:
    repository: ghcr.io/fortedigital/10x
    tag: 2.0.4                  # Updated by CI/CD

  extraEnv:
  - name: PORT
    value: "3000"
  - name: SKILLS_DIR
    value: "/app/skills"
  - name: FLOWCASE_ENDPOINT
    value: "https://forte.cvpartner.com/api/"

  envSecretName: "app-credentials"

auth:
  enabled: false
  tokens:
  - d4f88f6d9292c10cc3e21c4aad56d2be485db532b54fe961d738e1137d247823

ingress:
  enabled: true
  host: mcp10x.forteapps.net

Helm Chart Reference

Template Functions

forteapp.fullname

{{ include "forteapp.fullname" . }}
# Output: <release-name>

forteapp.labels

{{ include "forteapp.labels" . }}
# Output:
# app.kubernetes.io/name: forteapp
# app.kubernetes.io/instance: <release-name>
# app.kubernetes.io/version: <chart-version>
# app.kubernetes.io/managed-by: Helm

forteapp.selectorLabels

{{ include "forteapp.selectorLabels" . }}
# Output:
# app.kubernetes.io/name: forteapp
# app.kubernetes.io/instance: <release-name>

Deployment Specification

apiVersion: apps/v1
kind: Deployment
metadata:
  name: {{ include "forteapp.fullname" . }}
  labels:
    {{- include "forteapp.labels" . | nindent 4 }}
spec:
  replicas: {{ .Values.app.replicaCount }}
  selector:
    matchLabels:
      {{- include "forteapp.selectorLabels" . | nindent 6 }}
  template:
    metadata:
      annotations:
        policies.forteapps.io/auth: {{ .Values.auth.enabled | quote }}
      labels:
        {{- include "forteapp.selectorLabels" . | nindent 8 }}
    spec:
      containers:
      - name: app
        image: "{{ .Values.app.image.repository }}:{{ .Values.app.image.tag }}"
        imagePullPolicy: {{ .Values.app.image.pullPolicy }}
        ports:
        - name: http
          containerPort: {{ .Values.app.image.containerPort }}
        env:
        - name: NODE_ENV
          value: {{ .Values.app.nodeEnv | quote }}
        {{- with .Values.app.extraEnv }}
        {{- toYaml . | nindent 8 }}
        {{- end }}
        {{- if .Values.app.envSecretName }}
        envFrom:
        - secretRef:
            name: {{ .Values.app.envSecretName }}
        {{- end }}
        resources:
          {{- toYaml .Values.app.resources | nindent 10 }}
        securityContext:
          readOnlyRootFilesystem: true
          allowPrivilegeEscalation: false

IngressRoute Specification

apiVersion: traefik.io/v1alpha1
kind: IngressRoute
metadata:
  name: {{ include "forteapp.fullname" . }}
spec:
  entryPoints:
  - {{ .Values.ingress.entrypoint }}
  routes:
  - match: Host(`{{ .Values.ingress.host }}`)
    kind: Rule
    services:
    - name: {{ include "forteapp.fullname" . }}
      port: {{ .Values.service.port }}
  {{- if .Values.ingress.tls.enabled }}
  tls:
    secretName: {{ default .Release.Name .Values.ingress.tls.secretName }}-tls
  {{- end }}

Certificate Specification

apiVersion: cert-manager.io/v1
kind: Certificate
metadata:
  name: {{ include "forteapp.fullname" . }}-tls
spec:
  secretName: {{ default .Release.Name .Values.ingress.tls.secretName }}-tls
  issuerRef:
    name: {{ .Values.ingress.tls.clusterIssuer }}
    kind: ClusterIssuer
  dnsNames:
  - {{ .Values.ingress.host }}

ArgoCD Configuration

Application Manifest Schema

apiVersion: argoproj.io/v1alpha1
kind: Application
metadata:
  name: <app-name>
  namespace: argocd
  annotations:
    argocd.argoproj.io/sync-wave: "1"
    notifications.argoproj.io/subscribe.on-sync-succeeded.slack: ""
    notifications.argoproj.io/subscribe.on-sync-failed.slack: ""
    notifications.argoproj.io/subscribe.on-degraded.slack: ""
  labels:
    app.kubernetes.io/name: <app-name>
    app.kubernetes.io/part-of: apps
    app.kubernetes.io/managed-by: argocd
  finalizers:
  - resources-finalizer.argocd.argoproj.io

spec:
  project: default

  # Multi-source configuration
  sources:
  - repoURL: https://git.forteapps.net/Forte/forte-helm
    path: forteapp
    targetRevision: HEAD
    helm:
      valueFiles:
      - $values/<app-name>/values.yaml

  - repoURL: git@github.com:fortedigital/helm-prod-values.git
    targetRevision: HEAD
    ref: values

  destination:
    server: https://kubernetes.default.svc
    namespace: <app-name>

  syncPolicy:
    automated:
      prune: true
      selfHeal: true
      allowEmpty: false

    syncOptions:
    - CreateNamespace=true
    - Validate=true
    - ServerSideApply=true
    - Replace=false

    retry:
      limit: 5
      backoff:
        duration: 5s
        factor: 2
        maxDuration: 3m

  ignoreDifferences:
  - group: apps
    kind: Deployment
    jsonPointers:
    - /spec/replicas

Sync Waves

Wave Components Purpose
-1 Namespaces Create namespaces first
0 Kyverno Install policy engine
1 Cluster resources, infrastructure Base infrastructure
2+ Applications Business applications

Sync Options

Option Description
CreateNamespace=true Automatically create target namespace
Validate=true Validate resources before applying
ServerSideApply=true Use server-side apply (safer)
Replace=false Don't use kubectl replace
Prune=true Delete resources not in Git

Retry Policy

retry:
  limit: 5                 # Max retry attempts
  backoff:
    duration: 5s           # Initial backoff
    factor: 2              # Exponential factor
    maxDuration: 3m        # Max backoff time

Retry Schedule:

  1. 5 seconds
  2. 10 seconds
  3. 20 seconds
  4. 40 seconds
  5. 80 seconds (capped at 3 minutes)

Global Settings (argocd-cm)

Setting Value Purpose
application.resourceTrackingMethod annotation Track resources via annotations
timeout.reconciliation 60s Reconciliation interval
admin.enabled false Admin login disabled (SSO-only)
url https://argocd.forteapps.net External URL for ArgoCD UI

Git Submodule Disable: Set via configs.params (NOT repoServer.env — that causes strategic merge conflicts with chart's valueFrom entries):

configs:
  params:
    "reposerver.enable.git.submodule": "false"

This writes to argocd-cmd-params-cm ConfigMap, which the chart already reads via valueFrom. Submodules (e.g., shared-prompts) are not needed for K8s manifest generation.

Break-Glass Admin Access: Admin login is disabled (admin.enabled: false). The admin password remains in argocd-secret. To re-enable temporarily:

# Enable admin login
kubectl patch cm argocd-cm -n argocd -p '{"data":{"admin.enabled":"true"}}'
# Log in as admin, do what's needed, then disable again
kubectl patch cm argocd-cm -n argocd -p '{"data":{"admin.enabled":"false"}}'

ArgoCD picks up ConfigMap changes within the reconciliation timeout (60s). Note: ArgoCD will revert this on next sync — this is intentional (temporary access only).

OIDC Authentication (Keycloak):

configs:
  cm:
    oidc.config: |
      name: Forte SSO
      issuer: https://id.forteapps.net/realms/forte
      clientID: argocd
      clientSecret: $oidc.clientSecret
      requestedScopes: ["openid", "email", "profile"]
  rbacConfig:
    policy.csv: |
      g, ArgoCD Admins, role:admin
      g, ArgoCD Viewers, role:readonly
    # Deny users not in any declared KC group
    policy.default: ""
    scopes: '[groups]'

Access Control: Only users in Keycloak groups ArgoCD Admins or ArgoCD Viewers can access ArgoCD. Users not in either group are denied (empty policy.default). Assign users to groups in Keycloak admin console.

  • ArgoCD does NOT add openid implicitly — must include in requestedScopes
  • Do NOT add groups as a scope — the KC groups mapper emits the claim regardless
  • $oidc.clientSecret references the oidc.clientSecret key in argocd-secret
  • OIDC secret is synced by CronJob argocd-oidc-sync (see cluster-resources/argocd-oidc-secret-sync.yaml)
  • The CronJob bridges argocd-oidc-credentials (from KC registrar) → argocd-secret every 2 min
  • Safe for fresh deploys: no-ops if source secret doesn't exist yet

Ingress (Traefik + TLS):

server:
  ingress:
    enabled: true
    ingressClassName: traefik
    annotations:
      cert-manager.io/cluster-issuer: letsencrypt-prod
    tls: true
  extraArgs:
  - --insecure
configs:
  params:
    "server.insecure": true

TLS terminates at Traefik; ArgoCD runs in insecure mode behind the proxy.

Infrastructure Components

Homepage (Platform Dashboard)

Chart: jameswynn/homepage Namespace: homepage URL: https://start.forteapps.net

Platform dashboard that auto-discovers deployed apps via Kubernetes service annotations.

Discovery mechanism: Services annotated with gethomepage.dev/enabled: "true" appear in the dashboard. Apps not deployed = annotations absent = not shown. Fully dynamic per environment.

Annotated services:

Service Namespace Group Widget
gitea-http gitea DevOps gitea
argocd-server argocd DevOps argocd
keycloak keycloak Identity none
grafana monitoring Monitoring grafana
karpor-server karpor DevOps none

Adding a new app: Annotate the app's Service in its Helm values:

service:
  annotations:
    gethomepage.dev/enabled: "true"
    gethomepage.dev/name: "My App"
    gethomepage.dev/description: "What it does"
    gethomepage.dev/group: "GroupName"
    gethomepage.dev/icon: "icon-name"     # https://github.com/walkxcode/dashboard-icons
    gethomepage.dev/href: "https://myapp.forteapps.net"
    # Optional live widget:
    gethomepage.dev/widget.type: "myapp"
    gethomepage.dev/widget.url: "https://myapp.forteapps.net"
    # gethomepage.dev/widget.key: "{{HOMEPAGE_VAR_MYAPP_TOKEN}}"

Widget API credentials: Inject via env vars into the Homepage pod:

# In homepage-values.yaml per environment
env:
- name: HOMEPAGE_VAR_GRAFANA_TOKEN
  valueFrom:
    secretKeyRef:
      name: homepage-widget-credentials
      key: grafana-token

Then reference as gethomepage.dev/widget.key: "{{HOMEPAGE_VAR_GRAFANA_TOKEN}}".

Values files:

  • infra/values/base/homepage-values.yaml — RBAC, kubernetes mode, layout
  • infra/values/{env}/homepage-values.yaml — hostname per environment

Traefik

Chart: traefik/traefik Version: Latest Namespace: traefik

Configuration:

# infra/base/traefik-application.yaml
replicas: 2

service:
  type: LoadBalancer

ingressRoute:
  dashboard:
    enabled: false

ports:
  web:
    redirectTo: websecure  # HTTP → HTTPS redirect
  websecure:
    tls:
      enabled: true

Endpoints:

  • HTTP: :80 → Redirects to HTTPS
  • HTTPS: :443

Cert-Manager

Chart: jetstack/cert-manager Namespace: cert-manager

ClusterIssuer:

apiVersion: cert-manager.io/v1
kind: ClusterIssuer
metadata:
  name: letsencrypt-prod
spec:
  acme:
    server: https://acme-v02.api.letsencrypt.org/directory
    email: admin@forteapps.net
    privateKeySecretRef:
      name: letsencrypt-prod-key
    solvers:
    - http01:
        ingress:
          class: traefik

Kyverno

Chart: kyverno/kyverno Namespace: kyverno

Policies:

  • Secret cloner
  • Default namespace blocker
  • Bare pod cleaner
  • ReplicaSet cleaner
  • Deployment verifier
  • Auth sidecar injector

Sealed Secrets

Chart: sealed-secrets/sealed-secrets-controller Namespace: kube-system

Directory Structure: secrets/base/ contains all SealedSecrets with a kustomization.yaml. Per-cloud overlays in secrets/overlays/<cloud>/ reference the base via Kustomize. The ArgoCD secrets Application points to the active overlay (e.g., secrets/overlays/upc-dev), and infra/overlays/upc-prod patches the path to secrets/overlays/upc-prod.

To add cloud-specific secrets, create a new SealedSecret in the overlay directory and add it to the overlay's kustomization.yaml.

Public Certificate:

kubeseal --fetch-cert \
  --controller-name=sealed-secrets-controller \
  --controller-namespace=kube-system \
  > pub-cert.pem

Prometheus

Chart: prometheus-community/prometheus Namespace: monitoring

Configuration:

server:
  persistentVolume:
    enabled: true
    size: 10Gi

alertmanager:
  enabled: false

nodeExporter:
  enabled: true

kubeStateMetrics:
  enabled: true

Grafana

Chart: grafana/grafana Namespace: monitoring

Datasources:

  • Prometheus
  • Loki
  • Tempo

Ingress: Exposed via Traefik at https://grafana.forteapps.net with cert-manager TLS.

OIDC Authentication (Keycloak):

  • Uses grafana.ini.auth.generic_oauth with KC grafana client
  • Secret grafana-oidc-credentials synced by KC registrar, loaded via envFromSecrets
  • SSO-only mode: auth.disable_login_form: true + auth.generic_oauth.auto_login: true
  • Role mapping via JMESPath on resource_access.grafana.roles claim (requires KC client role mapper)
  • Roles: KC client roles Admin/Editor map to Grafana roles; default is Viewer

Loki

Chart: grafana/loki-stack Namespace: monitoring

Configuration:

loki:
  persistence:
    enabled: true
    size: 10Gi

promtail:
  enabled: false  # Using Fluent-Bit instead

Tempo

Chart: grafana/tempo Version: 1.24.4 Namespace: monitoring

Purpose: Distributed tracing backend receiving OTLP traces from Traefik and other instrumented services.

Configuration:

tempo:
  storage:
    trace:
      backend: local
      local:
        path: /var/tempo/traces
  receivers:
    otlp:
      protocols:
        grpc:
          endpoint: "0.0.0.0:4317"
        http:
          endpoint: "0.0.0.0:4318"

persistence:
  enabled: true
  size: 10Gi

Endpoints:

  • gRPC OTLP receiver: :4317
  • HTTP OTLP receiver: :4318
  • Query API: :3200

Grafana Integration:

  • Trace-to-logs correlation with Loki (by namespace, pod, container)
  • Trace-to-metrics correlation with Prometheus (by service name)
  • Service graph and node graph visualization

Fluent-Bit

Chart: fluent/fluent-bit Namespace: monitoring

Output: Loki

Gitea

Chart: gitea/gitea Version: 12.5.0 (app v1.25.4) Namespace: gitea

Purpose: Self-hosted Git repository hosting with pull requests, issues, CI/CD (Gitea Actions), container registry, and package registry.

Configuration:

# infra/base/gitea.yaml + infra/values/base/gitea-values.yaml
ingress:
  host: git.forteapps.net
  tls: cert-manager (letsencrypt-prod)

gitea:
  admin:
    existingSecret: gitea-credentials
  config:
    service:
      DISABLE_REGISTRATION: true
      ALLOW_ONLY_EXTERNAL_REGISTRATION: true
    actions:
      ENABLED: true
    packages:
      ENABLED: true
    metrics:
      ENABLED: true

postgresql:
  enabled: true
  persistence: 8Gi (upcloud-block-storage-maxiops)

Authentication: Keycloak OIDC via forte realm (client ID: gitea). Protocol mapper: email_verified hardcoded claim (true, boolean) on ID token, Access token, and Userinfo.

External User Sync: Disabled (cron.sync_external_users.ENABLED: false). This Gitea cron job is designed for LDAP and deactivates OIDC-only users because it cannot enumerate them — causing "Sign-in prohibited" errors after the sync runs.

Email Notifications: Enabled (ENABLE_NOTIFY_MAIL: true). SMTP credentials injected via gitea-smtp-secret using additionalConfigFromEnvs with GITEA__mailer__USER / GITEA__mailer__PASSWD environment variables.

Auto-Watch: Disabled (AUTO_WATCH_ON_CHANGES: false, AUTO_WATCH_NEW_REPOS: false). Prevents contributors from being auto-subscribed to repo notifications on push, reducing email noise from CI bots (e.g., ai-review PR comments). Users who were already watching before this change need to manually unwatch or switch to "Only participating".

Endpoints:

  • Web UI: https://git.forteapps.net
  • SSH: port 22 (ClusterIP)
  • Metrics: /metrics (Prometheus scrape)

Secrets:

  • gitea-credentials (SealedSecret) — admin password
  • gitea-oidc-credentials (registrar-managed) — OIDC client ID + secret
  • gitea-smtp-secret (SealedSecret) — SMTP username + password

Gitea Actions Runners

Chart: actions (from https://dl.gitea.com/charts) Namespace: gitea Sync Wave: 2 (deploys after Gitea)

Purpose: Act runners execute Gitea Actions CI/CD workflows. Deployed as a StatefulSet with a Docker-in-Docker sidecar for container-based job execution.

Configuration:

# infra/base/gitea-actions.yaml + infra/values/base/gitea-actions-values.yaml
replicaCount: 3

runner:
  labels:
    - "ubuntu-latest:docker://node:20-bookworm"
    - "ubuntu-22.04:docker://node:20-bookworm"
  existingSecret: gitea-runner-token

gitea:
  instance:
    url: http://gitea-http.gitea.svc.cluster.local:3000

dind:
  enabled: true  # Docker-in-Docker sidecar (privileged)

Resources:

Container CPU Request Memory Request CPU Limit Memory Limit
Runner 250m 256Mi 1 1Gi
DinD sidecar 250m 256Mi 1 1Gi

Secrets: gitea-runner-token (SealedSecret) containing token (instance-level runner registration token from /admin/runners)

Setup Steps:

  1. Get runner registration token from Gitea admin panel (/admin/runners)
  2. Fill in private/gitea-runner-token.yaml with the token
  3. Seal: kubeseal --format yaml < private/gitea-runner-token.yaml > secrets/gitea-runner-token-sealed.yaml
  4. Commit and push — ArgoCD deploys runners automatically

Verification:

  • kubectl get statefulset -n gitea — 3/3 runners ready
  • Gitea admin panel (/admin/runners) — runners show as Online
  • Create test workflow in .gitea/workflows/test.yml — job executes

Vaultwarden

Chart: guerzon/vaultwarden Version: 0.36.4 (app v1.36.0-alpine) Namespace: vaultwarden

Purpose: Self-hosted Bitwarden-compatible password manager.

Configuration:

# infra/overlays/upc-dev/vaultwarden/ + infra/values/
domain: "https://vaultwarden.forteapps.net"

ingress:
  enabled: true
  class: "traefik"
  tls: true
  tlsSecret: vaultwarden-tls
  hostname: bitwarden.forteapps.net
  additionalAnnotations:
    cert-manager.io/cluster-issuer: letsencrypt-prod

database:
  type: postgresql
  existingSecret: prod-db-creds

storage:
  data: 5Gi (ReadWriteOnce)
  attachments: 5Gi (ReadWriteOnce)

TLS: cert-manager auto-provisions Let's Encrypt certificate via letsencrypt-prod ClusterIssuer (same pattern as Gitea, Grafana, etc).

Endpoints:

  • Web UI: https://bitwarden.forteapps.net

Secrets:

  • prod-db-creds — PostgreSQL credentials + SMTP credentials
  • vaultwarden-tls — auto-managed by cert-manager

AI Code Review (ai-review)

Type: Gitea Actions workflow (.gitea/workflows/ai-review.yaml) Trigger: pull_request events (opened, synchronize) Runner: ubuntu-latest (container: nikitafilonov/ai-review:latest)

Purpose: Automated AI-powered code review on pull requests using Claude (Anthropic). Posts inline comments on changed lines and a PR summary comment highlighting infrastructure impact.

Architecture:

  • Uses xai-review Docker image
  • Shared configuration and prompts live in the shared-prompts Git submodule (→ Forte/ai-review-prompts)
  • Review mode: ONLY_ADDED_WITH_CONTEXT — reviews only new/changed lines plus surrounding context (token-efficient)
  • Agent mode: disabled (one-shot review, no multi-turn reasoning)
  • LLM: Claude Sonnet (claude-sonnet-4-20250514)

Shared Prompts Structure (submodule: Forte/ai-review-prompts):

shared-prompts/
  base/
    security.md          # org-wide security rules (all profiles)
  iac/
    .ai-review.yaml      # IaC/GitOps profile config
    inline.md            # inline review prompt
    summary.md           # PR summary prompt
  # future profiles: backend/, frontend/, etc.

Configuration (shared-prompts/iac/.ai-review.yaml):

llm:
  provider: CLAUDE
  model: claude-sonnet-4-20250514
vcs:
  provider: GITEA
review:
  mode: ONLY_ADDED_WITH_CONTEXT
agent:
  enabled: false
prompt:
  inline_prompt_files:            # concatenated in order
    - ./shared-prompts/base/security.md
    - ./shared-prompts/iac/inline.md
  summary_prompt_files:
    - ./shared-prompts/iac/summary.md
ignore:
  - "*.sealed.yaml"
  - "*.lock"
  - "docs/**"

Custom Prompts (IaC profile):

  • shared-prompts/base/security.md — org-wide security rules, concatenated before every inline review prompt
  • shared-prompts/iac/inline.md — IaC-specific inline review (YAML, Helm, K8s manifests, shell scripts), max 7 comments
  • shared-prompts/iac/summary.md — PR summary: affected services/namespaces, infrastructure impact, security flags

Prompt composition: ai-review does not support Jinja includes. Instead, list multiple files under inline_prompt_files / summary_prompt_files — they are concatenated in order with double newlines.

Adding a new profile: Create a new directory (e.g., backend/) with its own .ai-review.yaml, inline.md, and summary.md. The inline_prompt_files list should include base/security.md first, then the profile-specific prompt. Reference it in the consuming repo's workflow: AI_REVIEW_CONFIG_FILE_YAML=./shared-prompts/backend/.ai-review.yaml

Required Secrets (configure in Gitea repo or org settings):

Secret Purpose
ANTHROPIC_API_KEY Claude API key (from Anthropic console)
AI_REVIEW_TOKEN Gitea API token with write:repository + read:repository scopes (use a bot/service account)

Setup Steps:

  1. Create a Gitea bot/service account and generate an API token with write:repository + read:repository scopes
  2. Add AI_REVIEW_TOKEN secret in Gitea repo settings → Actions → Secrets
  3. Add ANTHROPIC_API_KEY secret with your Anthropic API key
  4. Ensure the shared-prompts submodule is initialized (git submodule update --init)
  5. Push the workflow file — it triggers automatically on PR creation/update

Verification:

  • Open a PR with infrastructure changes → workflow runs → inline comments + summary appear
  • Check Gitea Actions tab for workflow run status and logs
  • Monitor Anthropic usage dashboard for token consumption

Keycloak Browser Flow (IdP Auto-Redirect)

File: infra/values/base/keycloak-values.yaml (inside forte-realm.json)

The realm uses a custom browser authentication flow (browser-auto-idp) that skips the Keycloak login page and redirects directly to the Entra ID identity provider.

Flow executions:

Priority Authenticator Requirement Purpose
10 auth-cookie ALTERNATIVE Reuse existing session (no redirect)
20 identity-provider-redirector ALTERNATIVE Auto-redirect to forte-entra IdP

Key fields in realm JSON:

  • "browserFlow": "browser-auto-idp" — overrides the default browser flow at realm level
  • "authenticationFlows" — defines the custom flow with its executions
  • "authenticatorConfig" — sets defaultProvider: "forte-entra" on the redirector

Why custom flow: The default KC browser flow shows a username/password form with an IdP button. Since all authentication is via Entra ID, the custom flow eliminates this step. The auth-cookie execution preserves session reuse so returning users aren't redirected again.

Important: The forte-entra identity provider must exist in Keycloak (currently configured manually in the KC admin console). If the IdP alias changes, update the defaultProvider value in the realm JSON.

Keycloak Client Registrar

Type: CronJob (deployed via Keycloak Helm chart extraDeploy) Namespace: keycloak Schedule: */2 * * * * (every 2 minutes)

Purpose: Handles two responsibilities:

  1. Legacy sync — extracts secrets from Keycloak clients with k8s.secret.sync: "true" attribute (same as former PostSync syncer)
  2. Self-service registration — processes config Secrets (cloned by Kyverno) to register new OIDC clients and sync their credentials

How It Works:

Legacy path (existing clients like Gitea):

  1. Authenticates to Keycloak Admin API using admin credentials from keycloak-credentials secret
  2. Queries all clients in the forte realm
  3. Filters clients with k8s.secret.sync: "true" attribute
  4. For each matching client, retrieves the auto-generated secret via Keycloak Admin API
  5. Creates/updates a K8s Secret in the target namespace (from k8s.secret.namespace attribute)
  6. Always writes a central copy to the secrets namespace

Self-service path (new clients):

  1. Lists Secrets in keycloak namespace with label keycloak.forteapps.net/client-config=true
  2. For each config Secret, parses client.json and computes a config hash
  3. Skips if hash matches annotation and credential Secret already exists
  4. Creates or updates the Keycloak client via Admin API
  5. Fetches the generated client secret
  6. Upserts credential Secret in target namespace + central secrets namespace
  7. Annotates config Secret with sync status, config hash, and timestamp

Resources:

  • ServiceAccount: keycloak-client-registrar (namespace: keycloak)
  • ClusterRole: keycloak-client-registrar (secrets: get/list/create/update/patch; namespaces: get/list)
  • ClusterRoleBinding: keycloak-client-registrar
  • CronJob: keycloak-client-registrar

Kyverno Policy: keycloak-client-config-cloner — clones labeled Secrets from app namespaces to keycloak namespace (see Kyverno Policies)

Legacy Client Attributes (set in forte-realm.json):

Attribute Required Default Description
k8s.secret.sync Yes — Set to "true" to enable syncing
k8s.secret.namespace Yes — Target K8s namespace
k8s.secret.name Yes — Name of the K8s Secret
k8s.secret.client-id-key No client-id Field name for client ID in the Secret
k8s.secret.client-secret-key No client-secret Field name for client secret in the Secret

Self-Service Config Secret Schema:

apiVersion: v1
kind: Secret
metadata:
  name: keycloak-client-<app>
  namespace: <app-namespace>
  labels:
    keycloak.forteapps.net/client-config: "true"
stringData:
  client.json: |
    {
      "clientId": "<app>",
      "name": "<App Name>",
      "redirectUris": ["https://<app>.forteapps.net/*"],
      "webOrigins": ["https://<app>.forteapps.net"],
      "defaultClientScopes": ["openid", "email", "profile"],
      "protocolMappers": [],
      "secret": {
        "namespace": "<app-namespace>",
        "name": "<app>-oidc-credentials",
        "keys": { "clientId": "client-id", "clientSecret": "client-secret" }
      }
    }

Created Credential Secret Format:

apiVersion: v1
kind: Secret
metadata:
  name: <target-name>
  namespace: <target-namespace>
  labels:
    app.kubernetes.io/managed-by: keycloak-client-registrar
type: Opaque
data:
  <client-id-key>: <base64-encoded client ID>
  <client-secret-key>: <base64-encoded client secret>

Config Secret Annotations (set by registrar):

Annotation Description
keycloak.forteapps.net/config-hash SHA-256 hash of client.json for change detection
keycloak.forteapps.net/sync-status synced or error
keycloak.forteapps.net/last-sync ISO 8601 timestamp of last successful sync

Verification:

# Check CronJob status
kubectl get cronjobs -n keycloak

# View latest registrar logs
kubectl logs -n keycloak job/$(kubectl get jobs -n keycloak --sort-by=.metadata.creationTimestamp -o jsonpath='{.items[-1].metadata.name}')

# Verify created secret
kubectl get secret <name> -n <namespace> -o yaml

# Check config Secret annotations (self-service)
kubectl get secret keycloak-client-<app> -n keycloak -o jsonpath='{.metadata.annotations}'

See: Developer Guide - Adding a New Keycloak Client

Karpor

Chart: karpor from https://kusionstack.github.io/charts Version: 0.7.6 (app v0.6.4) Namespace: karpor Sync Wave: 1

Purpose: Kubernetes visualization and intelligence tool. Provides cross-cluster resource search, compliance checking, and topology visualization. Gives platform engineers a unified view of all cluster resources and their relationships.

Architecture (4 components):

  • Server — main Karpor API/UI (port 7443)
  • Syncer — syncs cluster state into the search index
  • ElasticSearch — search backend for resource indexing
  • etcd — persistent key-value store (10Gi PVC)

Configuration (infra/values/base/karpor-values.yaml):

  • namespaceEnabled: false — ArgoCD manages namespace creation
  • Default resource limits tuned for small clusters
  • ElasticSearch: 2 CPU / 4Gi memory (the heaviest component)
  • AI features available but not enabled (requires server.ai.authToken + backend config)

Access: Port-forward to reach the UI:

kubectl port-forward svc/karpor-release-server -n karpor 7443:7443
# Open https://localhost:7443

Renovate

Chart: renovate (OCI: ghcr.io/renovatebot/charts) Version: 46.109.0 (app v43.113.0) Namespace: renovate Sync Wave: 2

Purpose: Automated dependency update bot. Runs as a CronJob that scans Gitea repositories for outdated dependencies and creates pull requests with updates.

Configuration:

# infra/base/renovate.yaml + infra/values/base/renovate-values.yaml
cronjob:
  schedule: "@daily"
  concurrencyPolicy: Forbid

renovate:
  config:
    platform: gitea
    endpoint: https://git.forteapps.net
    autodiscover: true
    gitAuthor: "Renovate Bot <renovate@forteapps.net>"
    packageRules:
      - matchRepositories: ["**/10x"]
        assignees: ["edvard.unsvag"]
        reviewers: ["edvard.unsvag"]
      - matchRepositories: ["**/auth-sidecar"]
        assignees: ["danijel.simeunovic"]
        reviewers: ["danijel.simeunovic"]
      - matchRepositories: ["**/forte-helm"]
        assignees: ["danijel.simeunovic"]
        reviewers: ["danijel.simeunovic"]

resources:
  requests: { cpu: 500m, memory: 1Gi }
  limits: { cpu: "2", memory: 4Gi }

Note: Assignees and reviewers are only applied at PR creation time. Existing PRs must be closed and recreated for new assignment rules to take effect.

Secrets: renovate-env (SealedSecret in secrets namespace, cloned by Kyverno) containing:

  • RENOVATE_TOKEN — Gitea PAT with repo write + issue write permissions
  • RENOVATE_GITHUB_COM_TOKEN — GitHub PAT (public_repo read-only) for changelog fetching

Setup Steps:

  1. Fill in private/renovate-env.yaml with tokens
  2. Seal: kubeseal --format yaml < private/renovate-env.yaml > secrets/renovate-env-sealed.yaml
  3. Commit and push — ArgoCD deploys the CronJob, Kyverno clones the secret

Verification:

  • kubectl get cronjob -n renovate — CronJob exists
  • kubectl create job --from=cronjob/renovate renovate-test -n renovate — manual trigger
  • kubectl logs -n renovate job/renovate-test — check logs

Kyverno Policies

Secret Cloner

File: cluster-resources/policies/secret-cloner.yaml

Purpose: Automatically clone secrets from secrets namespace to new namespaces

apiVersion: kyverno.io/v1
kind: ClusterPolicy
metadata:
  name: sync-secret-with-multi-clone
spec:
  rules:
  - name: clone-secret
    match:
      any:
      - resources:
          kinds:
          - Namespace
    generate:
      apiVersion: v1
      kind: Secret
      name: "{{ request.object.metadata.name }}"
      namespace: "{{ request.object.metadata.name }}"
      synchronize: true
      clone:
        namespace: secrets
        name: shared-credentials

Label Requirement: Secrets must have allowedToBeCloned: "true"

Keycloak Client Config Cloner

File: cluster-resources/policies/keycloak-client-cloner.yaml

Purpose: Clones Secrets labeled keycloak.forteapps.net/client-config: "true" from app namespaces to the keycloak namespace. This allows apps to declare their OIDC client configuration in their own namespace, which the Keycloak Client Registrar then processes.

Trigger: Any Secret with label keycloak.forteapps.net/client-config: "true" created outside the keycloak namespace.

Behavior:

  • Generates a copy of the Secret in the keycloak namespace with the same name
  • Adds source tracking annotations (keycloak.forteapps.net/source-namespace, keycloak.forteapps.net/source-name)
  • synchronize: true — changes to the source Secret are reflected in the clone

Default Namespace Blocker

File: cluster-resources/policies/default-ns-blocker.yaml

Purpose: Prevent resources from being created in default namespace

apiVersion: kyverno.io/v1
kind: ClusterPolicy
metadata:
  name: disallow-default-namespace
spec:
  validationFailureAction: enforce
  rules:
  - name: validate-namespace
    match:
      any:
      - resources:
          kinds:
          - Pod
          - Deployment
          - Service
    validate:
      message: "Using 'default' namespace is not allowed"
      pattern:
        metadata:
          namespace: "!default"

Bare Pod Cleaner

File: cluster-resources/policies/bare-pod-cleaner.yaml

Purpose: Delete pods without ownerReferences (not managed by Deployment/StatefulSet)

apiVersion: kyverno.io/v1
kind: ClusterPolicy
metadata:
  name: cleanup-bare-pods
spec:
  rules:
  - name: delete-bare-pod
    match:
      any:
      - resources:
          kinds:
          - Pod
    preconditions:
      all:
      - key: "{{ request.object.metadata.ownerReferences[] || '' }}"
        operator: Equals
        value: ""
    validate:
      message: "Bare pods (without controllers) are not allowed"
      deny: {}

Auth Sidecar Injector

File: cluster-resources/policies/auth-sidecar-injector.yaml

Purpose: Automatically inject authentication sidecar into pods with authentication enabled

Rules: 6 rules in the policy

  1. generate-auth-tokens-secret - Creates Secret for token mode
  2. generate-auth-oidc-secret - Creates Secret for OIDC mode
  3. inject-sidecar-token - Injects auth sidecar for token mode
  4. inject-sidecar-oidc - Injects auth sidecar for OIDC mode
  5. inject-sidecar-mcp - Injects auth sidecar for MCP OAuth mode (RFC 9728 / RFC 7591)
  6. generate-auth-network-policy - Creates NetworkPolicy to restrict ingress

Trigger Annotation

policies.forteapps.io/auth: "true"

Authentication Modes

Token Mode (default):

# Annotations
policies.forteapps.io/auth: "true"
policies.forteapps.io/auth-type: "token"
policies.forteapps.io/auth-token-secret-name: "auth-tokens"
policies.forteapps.io/auth-upstream-url: "http://localhost:3000"

# Optional customization
policies.forteapps.io/auth-image: "ghcr.io/fortedigital/auth-sidecar"
policies.forteapps.io/auth-image-version: "latest"

OIDC Mode:

# Annotations (required)
policies.forteapps.io/auth: "true"
policies.forteapps.io/auth-type: "oidc"
policies.forteapps.io/auth-oidc-authority: "https://auth.example.com/realms/master"
policies.forteapps.io/auth-oidc-client-id: "myapp"

# Optional annotations
policies.forteapps.io/auth-oidc-callback-path: "/auth/callback"
policies.forteapps.io/auth-oidc-scopes: "openid,profile,email"
policies.forteapps.io/auth-upstream-url: "http://localhost:3000"
policies.forteapps.io/auth-image: "ghcr.io/fortedigital/auth-sidecar"
policies.forteapps.io/auth-image-version: "latest"

MCP Mode (OAuth 2.0 for MCP servers, implements RFC 9728 / RFC 7591):

# Annotations (required)
policies.forteapps.io/auth: "true"
policies.forteapps.io/auth-type: "mcp"
policies.forteapps.io/auth-mcp-resource: "https://mcp.example.com"
policies.forteapps.io/auth-mcp-authority: "https://auth.example.com"

# Optional annotations
policies.forteapps.io/auth-mcp-scopes: "read,write"
policies.forteapps.io/auth-upstream-url: "http://localhost:3000"
policies.forteapps.io/auth-log-level: "info"
policies.forteapps.io/auth-image: "ghcr.io/fortedigital/auth-sidecar"
policies.forteapps.io/auth-image-version: "latest"

Sidecar Container Specification

Token Mode:

name: authn
image: ghcr.io/fortedigital/auth-sidecar:latest
ports:
- containerPort: 8080
  name: auth
  protocol: TCP
env:
- name: AUTH_MODE
  value: "token"
- name: AUTH_LISTEN_ADDR
  value: ":8080"
- name: AUTH_UPSTREAM_URL
  value: "http://localhost:3000"
- name: AUTH_TOKEN_FILE
  value: "/etc/auth/tokens"
volumeMounts:
- name: auth-tokens
  mountPath: /etc/auth
  readOnly: true
resources:
  requests:
    cpu: 10m
    memory: 32Mi
  limits:
    cpu: 50m
    memory: 64Mi
securityContext:
  allowPrivilegeEscalation: false
  readOnlyRootFilesystem: true
  capabilities:
    drop: [ALL]

OIDC Mode:

name: authn
image: ghcr.io/fortedigital/auth-sidecar:latest
ports:
- containerPort: 8080
  name: auth
  protocol: TCP
env:
- name: AUTH_MODE
  value: "oidc"
- name: AUTH_LISTEN_ADDR
  value: ":8080"
- name: AUTH_UPSTREAM_URL
  value: "http://localhost:3000"
- name: AUTH_OIDC_AUTHORITY
  value: "https://auth.example.com/realms/master"
- name: AUTH_OIDC_CLIENT_ID
  value: "myapp"
- name: AUTH_OIDC_CALLBACK_PATH
  value: "/auth/callback"
- name: AUTH_OIDC_SCOPES
  value: "openid,profile,email"
- name: AUTH_OIDC_COOKIE_SECRET
  valueFrom:
    secretKeyRef:
      name: auth-oidc
      key: cookie-secret
- name: AUTH_OIDC_CLIENT_SECRET
  valueFrom:
    secretKeyRef:
      name: auth-oidc
      key: client-secret
resources:
  requests:
    cpu: 10m
    memory: 32Mi
  limits:
    cpu: 50m
    memory: 64Mi
securityContext:
  allowPrivilegeEscalation: false
  readOnlyRootFilesystem: true
  capabilities:
    drop: [ALL]

MCP Mode:

name: authn
image: ghcr.io/fortedigital/auth-sidecar:latest
ports:
- containerPort: 8080
  name: auth
  protocol: TCP
env:
- name: AUTH_MODE
  value: "mcp"
- name: AUTH_LISTEN_ADDR
  value: ":8080"
- name: AUTH_LOG_LEVEL
  value: "info"
- name: AUTH_UPSTREAM_URL
  value: "http://localhost:3000"
- name: AUTH_MCP_RESOURCE
  value: "https://mcp.example.com"
- name: AUTH_MCP_AUTHORIZATION_SERVERS
  value: "https://auth.example.com"
- name: AUTH_MCP_SCOPES_SUPPORTED
  value: "read,write"
resources:
  requests:
    cpu: 10m
    memory: 32Mi
  limits:
    cpu: 50m
    memory: 64Mi
securityContext:
  allowPrivilegeEscalation: false
  readOnlyRootFilesystem: true
  capabilities:
    drop: [ALL]

Generated Resources

Secret (Token Mode):

apiVersion: v1
kind: Secret
metadata:
  name: auth-tokens
  namespace: <app-namespace>
  labels:
    app.kubernetes.io/managed-by: kyverno
    app.kubernetes.io/created-by: inject-auth-sidecar
type: Opaque
data: {}  # Populated by Helm chart

Secret (OIDC Mode):

apiVersion: v1
kind: Secret
metadata:
  name: auth-oidc
  namespace: <app-namespace>
  labels:
    app.kubernetes.io/managed-by: kyverno
    app.kubernetes.io/created-by: inject-auth-sidecar
type: Opaque
data:
  client-secret: <base64>
  cookie-secret: <base64>

NetworkPolicy:

apiVersion: networking.k8s.io/v1
kind: NetworkPolicy
metadata:
  name: <pod-name>-auth-ingress
  namespace: <app-namespace>
  labels:
    app.kubernetes.io/managed-by: kyverno
    app.kubernetes.io/created-by: inject-auth-sidecar
spec:
  podSelector:
    matchLabels: <pod-labels>
  policyTypes:
  - Ingress
  ingress:
  - ports:
    - port: 8080
      protocol: TCP

Excluded Namespaces

The policy does NOT apply to:

  • kube-system
  • kyverno
  • argocd
  • cert-manager
  • monitoring

Health Checks

readinessProbe:
  httpGet:
    path: /healthz
    port: 8080
  initialDelaySeconds: 2
  periodSeconds: 5

livenessProbe:
  httpGet:
    path: /healthz
    port: 8080
  initialDelaySeconds: 5
  periodSeconds: 10

Request Flow

External Request → Traefik
    ↓
Service (port 8080)
    ↓
Pod: Auth Sidecar (port 8080)
    ├─ Validate credentials
    │  • Token mode: Check Bearer token
    │  • OIDC mode: Validate session or redirect to IdP
    │  • MCP mode: OAuth 2.0 via RFC 9728 discovery / RFC 7591 dynamic registration
    ↓
Forward to Application (localhost:3000)
    ↓
Application processes request

Forwarded Headers

After successful authentication, the sidecar injects user identity as HTTP headers before forwarding the request to the application container:

Header Description Auth Modes
X-Auth-User Username or display name Token, OIDC, MCP
X-Auth-Email User email address OIDC
X-Auth-Subject OIDC sub claim (stable user ID) OIDC, MCP
X-Auth-Groups Comma-separated group memberships OIDC (if groups scope)
X-Auth-Token The validated access token All modes

These headers are trustworthy because the auto-generated NetworkPolicy restricts pod ingress to the sidecar port only — external traffic cannot reach the application container directly, so headers cannot be spoofed.

Applications should read these headers to obtain authenticated user information (e.g. for display, authorisation decisions, or audit logging) instead of implementing their own authentication.

See: Developer Guide - Accessing Authenticated User Information for code examples.

Configuration Reference

Environment Variables

Common environment variables used across applications:

Variable Purpose Example
NODE_ENV Node.js environment production
PORT Application port 3000
DB_HOST Database host postgres
DB_PORT Database port 5432
DB_USER Database user app_user
DB_NAME Database name app_db
DB_PASSWORD Database password From secret
API_KEY External API key From secret

Resource Limits

Recommended resource allocation:

Application Type CPU Request Memory Request CPU Limit Memory Limit
Lightweight API 100m 128Mi 500m 512Mi
Standard Web App 200m 256Mi 1000m 1Gi
Heavy Processing 500m 512Mi 2000m 2Gi
Database 250m 256Mi 1000m 1Gi

Storage Classes

Storage classes are cloud-specific and configured in per-cluster value overrides (infra/values/{cluster}/gitea-values.yaml):

Cloud Storage Class Driver
UpCloud upcloud-block-storage-maxiops UpCloud CSI
AWS EKS gp3 EBS CSI
Azure AKS managed-csi-premium Azure Disk CSI
GCP GKE premium-rwo PD CSI 
# Example: base values omit storageClass (set in per-cluster overlay)
persistence:
  enabled: true
  accessMode: ReadWriteOnce
  size: 5Gi
  # storageClass set by infra/values/{cluster}/gitea-values.yaml

API Endpoints

ArgoCD API

# Server
https://argocd.127.0.0.1.nip.io

# Applications endpoint
GET /api/v1/applications

# Application details
GET /api/v1/applications/{name}

# Sync application
POST /api/v1/applications/{name}/sync

Prometheus API

# Query endpoint
GET /api/v1/query?query={promql}

# Query range
GET /api/v1/query_range?query={promql}&start={time}&end={time}&step={duration}

# Metrics
GET /api/v1/label/__name__/values

Tempo API

# Search traces
GET /api/search?q={traceql}

# Get trace by ID
GET /api/traces/{traceID}

# Service tag values
GET /api/v2/search/tag/resource.service.name/values

Loki API

# Query logs
GET /loki/api/v1/query?query={logql}

# Query range
GET /loki/api/v1/query_range?query={logql}&start={time}&end={time}

# Push logs
POST /loki/api/v1/push

Cloud Overlay Pattern

Overview

Cloud-specific configuration (StorageClass, LoadBalancer annotations, pricing models, etc.) lives in per-cloud overlay value files, not in base/. Adding a new cloud provider only requires a new overlay directory — no base changes.

Supported Clouds

Cloud Dev overlay Prod overlay StorageClass LB type
UpCloud upc-dev upc-prod upcloud-block-storage-maxiops UpCloud LB (proxy protocol v2)
Azure AKS aks-dev aks-prod managed-csi-premium Azure LB
AWS EKS eks-dev eks-prod gp3 AWS NLB (proxy protocol)
GCP GKE gke-dev gke-prod premium-rwo GCP NEG

Bootstrap any cluster with: ./bootstrap.sh <cluster> (e.g., ./bootstrap.sh aks-dev)

How It Works

Each ArgoCD Application uses multi-source Helm values with two value files:

# infra/base/gitea.yaml (example)
helm:
  valueFiles:
  - $values/infra/values/base/gitea-values.yaml      # [0] cloud-agnostic
  - $values/infra/values/upc-dev/gitea-values.yaml    # [1] cloud-specific (default: upc-dev)

The upc-prod Kustomize overlay patches index [1] to swap the cloud-specific file:

# infra/overlays/upc-prod/kustomization.yaml
- target:
    kind: Application
    name: gitea
  patch: |
    - op: replace
      path: /spec/sources/0/helm/valueFiles/1
      value: $values/infra/values/upc-prod/gitea-values.yaml

Components Using Cloud Overlays

Component Cloud-specific config Overlay value file
Traefik LB annotations, proxy protocol IPs traefik-values.yaml
Keycloak Hostname, TLS settings keycloak-values.yaml
Grafana Hostname, datasource URLs grafana-values.yaml
Gitea StorageClass (persistence + PostgreSQL) gitea-values.yaml
OpenCost Custom pricing model (CPU/RAM/storage rates) opencost-values.yaml

Backup CronJob

The gitea-backup CronJob uses a generic s3 alias for minio/mc. The actual endpoint and credentials come from the gitea-backup-s3 Sealed Secret, which is per-cloud. Reference scripts for different cloud providers are in scripts/backup/:

Script Provider Tool
s3-minio.sh S3-compatible (UpCloud, MinIO, Wasabi) minio/mc
aws-s3.sh AWS S3 aws CLI
azure-blob.sh Azure Blob Storage az CLI
gcp-gcs.sh GCP Cloud Storage gsutil

Adding a New Cloud Provider

To add support for a new cloud (e.g., oci-dev for Oracle Cloud):

  1. Cluster config: clusters/oci-dev.yaml — clusterName, domain, trustedIPs, cloudProvider
  2. Overlay value files in infra/values/oci-dev/:
    • traefik-values.yaml — LB annotations, proxy protocol config
    • keycloak-values.yaml — hostname
    • grafana-values.yaml — hostname
    • gitea-values.yaml — storageClass for persistence + PostgreSQL
    • opencost-values.yaml — pricing model or cloud billing integration
  3. Kustomize overlay: infra/overlays/oci-dev/kustomization.yaml — patch valueFiles[1] for each Application
  4. App-of-apps: _app-of-apps-oci-dev.yaml — points to infra/overlays/oci-dev
  5. Secrets overlay: secrets/overlays/oci-dev/kustomization.yaml — references ../../base, add cloud-specific SealedSecrets if needed
  6. Secrets patch: Add patch to infra/overlays/oci-dev/kustomization.yaml to swap secrets path to secrets/overlays/oci-dev
  7. Bootstrap: ./bootstrap.sh oci-dev

Glossary

Terms

App-of-Apps: ArgoCD pattern where a parent Application manages child Applications

GitOps: Operations approach where Git is the single source of truth

IngressRoute: Traefik CRD for routing external traffic to services

Multi-Source: ArgoCD feature allowing multiple Git sources per Application

SealedSecret: Encrypted secret that can be safely stored in Git

Sync Wave: Ordered deployment using annotations

Self-Heal: ArgoCD automatically reverts manual cluster changes

Prune: Automatically delete resources removed from Git

Annotations Reference

ArgoCD Annotations

# Sync wave (deployment order)
argocd.argoproj.io/sync-wave: "1"

# Refresh application
argocd.argoproj.io/refresh: "hard"

# Compare options
argocd.argoproj.io/compare-options: IgnoreExtraneous

# Sync options per resource
argocd.argoproj.io/sync-options: Prune=false

Kyverno Annotations

# Exclude from policy
policies.kyverno.io/exclude: "true"

# Severity
policies.kyverno.io/severity: high

Custom Annotations

# Authentication enabled
policies.forteapps.io/auth: "true"

# OIDC configuration
policies.forteapps.io/auth-oidc-authority: "https://..."
policies.forteapps.io/auth-oidc-client-id: "client-id"

Labels Reference

Standard Labels

# Application name
app.kubernetes.io/name: myapp

# Application instance
app.kubernetes.io/instance: myapp

# Application version
app.kubernetes.io/version: "1.0.0"

# Component type
app.kubernetes.io/component: frontend

# Part of larger application
app.kubernetes.io/part-of: ecommerce

# Managed by
app.kubernetes.io/managed-by: argocd

Custom Labels

# Allow secret cloning
allowedToBeCloned: "true"

# Environment
environment: production

# Team ownership
team: platform

Version Matrix

Component Versions

Component Version Chart Version
ArgoCD 2.9.0+ Latest
Traefik 2.10.0+ Latest
Cert-Manager 1.13.0+ Latest
Kyverno 1.10.0+ Latest
Sealed Secrets 0.24.0+ Latest
Prometheus 2.47.0+ Latest
Grafana 10.0.0+ Latest
Loki 2.9.0+ Latest
Tempo 2.6.0+ 1.24.4
Fluent-Bit 2.1.0+ Latest
Gitea 1.25.4 12.5.0
Gitea Act Runner Latest Latest
Renovate v43.113.0 46.109.0
PostgreSQL 16-alpine N/A
Trivy Latest Latest

Kubernetes Compatibility

  • Minimum: 1.24+
  • Tested: 1.28+
  • Recommended: Latest stable

Last Updated: 2026-04-22 Maintained By: Platform Team Version: 1.0.0

Reference in New Issue View Git Blame Copy Permalink