simstudioai--sim/helm/sim/examples/values-gcp.yaml

# values-gcp.yaml
#
# When to use: Google Kubernetes Engine (GKE). Configures Persistent Disk
# storage, Google Cloud Load Balancer with managed certs, Workload Identity
# annotations, and Tesla T4/V100 GPU node selectors for Ollama.
#
# Prerequisites:
#   - GKE cluster (Kubernetes 1.25+)
#   - Workload Identity enabled (for IAM-bound ServiceAccount)
#   - (optional) Managed certificate or cert-manager for TLS
#   - (optional) GPU node pool with accelerator labels for Ollama
#
# Install:
#   helm install sim ./helm/sim \
#     --namespace sim --create-namespace \
#     --values ./helm/sim/examples/values-gcp.yaml \
#     --set app.env.BETTER_AUTH_SECRET="$BETTER_AUTH_SECRET" \
#     --set app.env.ENCRYPTION_KEY="$ENCRYPTION_KEY" \
#     --set app.env.INTERNAL_API_SECRET="$INTERNAL_API_SECRET" \
#     --set app.env.CRON_SECRET="$CRON_SECRET" \
#     --set postgresql.auth.password="$POSTGRES_PASSWORD"
#
# Choosing a secret strategy
# --------------------------
# Pick one of three modes — see helm/sim/README.md for details.
#
#   1. Inline (default below): set values directly in app.env. The chart
#      writes every key into a chart-managed Secret. Simplest, but stores
#      plaintext in your values file.
#
#   2. existingSecret: pre-create a Kubernetes Secret out-of-band and
#      reference it via app.secrets.existingSecret.name. You are
#      responsible for ensuring the Secret contains every key your app
#      needs (BETTER_AUTH_SECRET, ENCRYPTION_KEY, INTERNAL_API_SECRET,
#      OPENAI_API_KEY, ...). The chart-managed Secret is not rendered in
#      this mode, so anything missing from your Secret will be empty at
#      runtime.
#
#   3. External Secrets Operator (ESO): sync from Google Secret Manager
#      via the External Secrets Operator. Map every key via
#      externalSecrets.remoteRefs.app.<KEY>. The chart fails template
#      rendering if a key is set in app.env but not mapped — see the
#      commented ESO block at the bottom of this file.

# Global configuration
global:
  imageRegistry: "ghcr.io"
  storageClass: "standard-rwo"

# Main application
app:
  enabled: true
  replicaCount: 2

  # Node selector for application pods
  # Uncomment and customize based on your GKE node labels:
  # nodeSelector:
  #   cloud.google.com/gke-nodepool: "default-pool"
  
  resources:
    limits:
      memory: "4Gi"
      cpu: "2000m"
    requests:
      memory: "2Gi"
      cpu: "1000m"
  
  # Production URLs (REQUIRED - update with your actual domain names)
  env:
    NEXT_PUBLIC_APP_URL: "https://simstudio.acme.com"
    BETTER_AUTH_URL: "https://simstudio.acme.com"
    # SOCKET_SERVER_URL is auto-detected (uses internal service http://sim-realtime:3002)
    NEXT_PUBLIC_SOCKET_URL: "https://simstudio-ws.acme.com"  # Public WebSocket URL for browsers
    
    # Security settings (REQUIRED - replace with your own secure secrets)
    # Generate using: openssl rand -hex 32
    BETTER_AUTH_SECRET: "your-secure-production-auth-secret-here"
    ENCRYPTION_KEY: "your-secure-production-encryption-key-here"
    INTERNAL_API_SECRET: "your-secure-production-internal-api-secret-here"
    CRON_SECRET: "your-secure-production-cron-secret-here"

    # Optional: API Key Encryption (RECOMMENDED for production)
    # Generate 64-character hex string using: openssl rand -hex 32
    API_ENCRYPTION_KEY: "your-64-char-hex-api-encryption-key-here"  # Optional but recommended
    
    NODE_ENV: "production"
    NEXT_TELEMETRY_DISABLED: "1"
    
    # GCP-specific environment variables
    GOOGLE_CLOUD_PROJECT: "your-project-id"
    GOOGLE_CLOUD_REGION: "us-central1"

# Realtime service
realtime:
  enabled: true
  replicaCount: 2

  # Node selector for realtime pods
  # Uncomment and customize based on your GKE node labels:
  # nodeSelector:
  #   cloud.google.com/gke-nodepool: "default-pool"
  
  resources:
    limits:
      memory: "4Gi"
      cpu: "1000m"
    requests:
      memory: "2Gi"
      cpu: "500m"
  
  env:
    NEXT_PUBLIC_APP_URL: "https://simstudio.acme.com"
    BETTER_AUTH_URL: "https://simstudio.acme.com"
    BETTER_AUTH_SECRET: "your-secure-production-auth-secret-here"
    ALLOWED_ORIGINS: "https://simstudio.acme.com"
    NODE_ENV: "production"

# Database migrations
migrations:
  enabled: true
  
  resources:
    limits:
      memory: "2Gi"
      cpu: "1000m"
    requests:
      memory: "1Gi"
      cpu: "500m"

# PostgreSQL database
postgresql:
  enabled: true

  # Node selector for database pods
  # Uncomment and customize (recommended: memory-optimized machine types):
  # nodeSelector:
  #   cloud.google.com/gke-nodepool: "database-pool"
  
  # Database authentication (REQUIRED - set secure credentials)
  auth:
    username: postgres
    password: "your-secure-postgres-password"
    database: simstudio
  
  # Resource allocation optimized for GKE
  resources:
    limits:
      memory: "4Gi"
      cpu: "2000m"
    requests:
      memory: "2Gi"
      cpu: "1000m"
  
  # Persistent storage using Google Cloud Persistent Disk
  persistence:
    enabled: true
    storageClass: "standard-rwo"
    size: 50Gi
    accessModes:
      - ReadWriteOnce
  
  # SSL/TLS configuration (requires cert-manager to be installed)
  tls:
    enabled: false  # Set to true if cert-manager is installed
    certificatesSecret: postgres-tls-secret
  
  # PostgreSQL performance tuning for GCP infrastructure
  config:
    maxConnections: 1000
    sharedBuffers: "2GB"
    maxWalSize: "8GB"
    minWalSize: "160MB"

# Ollama AI models with GPU acceleration (GCP GPU instances)
# Set ollama.enabled: false if you don't need local AI models
ollama:
  enabled: false
  replicaCount: 1

  # GPU node targeting - uncomment and customize for GPU node pools
  # Recommended: T4 or V100 GPU instances
  # nodeSelector:
  #   cloud.google.com/gke-nodepool: "gpu-pool"
  #   cloud.google.com/gke-accelerator: "nvidia-tesla-t4"
  
  tolerations:
    - key: "nvidia.com/gpu"
      operator: "Equal"
      value: "present"
      effect: "NoSchedule"
  
  # GPU resource allocation for AI model serving
  gpu:
    enabled: true
    count: 1
  
  resources:
    limits:
      memory: "16Gi"
      cpu: "4000m"
      nvidia.com/gpu: "1"
    requests:
      memory: "8Gi"
      cpu: "2000m"
  
  # High-performance SSD storage for AI models
  persistence:
    enabled: true
    storageClass: "premium-rwo"
    size: 100Gi
    accessModes:
      - ReadWriteOnce
  
  env:
    NVIDIA_DRIVER_CAPABILITIES: "all"
    OLLAMA_LOAD_TIMEOUT: "-1"
    OLLAMA_KEEP_ALIVE: "-1"
    OLLAMA_DEBUG: "1"

# Ingress using Google Cloud Load Balancer
ingress:
  enabled: true
  className: gce
  
  annotations:
    kubernetes.io/ingress.class: gce
    kubernetes.io/ingress.global-static-ip-name: "simstudio-ip"
    networking.gke.io/managed-certificates: "simstudio-ssl-cert"
    kubernetes.io/ingress.allow-http: "false"
  
  # Main application
  app:
    host: simstudio.acme.com
    paths:
      - path: /
        pathType: Prefix
  
  # Realtime service
  realtime:
    host: simstudio-ws.acme.com
    paths:
      - path: /
        pathType: Prefix
  
  # TLS configuration
  tls:
    enabled: true
    secretName: simstudio-tls-secret

# Pod disruption budget for high availability
podDisruptionBudget:
  enabled: true
  minAvailable: null
  maxUnavailable: 1  
  unhealthyPodEvictionPolicy: AlwaysAllow 
# Network policies
networkPolicy:
  enabled: true

# Pod anti-affinity for high availability across GCP zones
affinity:
  podAntiAffinity:
    preferredDuringSchedulingIgnoredDuringExecution:
      - weight: 100
        podAffinityTerm:
          labelSelector:
            matchExpressions:
              - key: app.kubernetes.io/name
                operator: In
                values: ["simstudio"]
          topologyKey: kubernetes.io/hostname
      - weight: 50
        podAffinityTerm:
          labelSelector:
            matchExpressions:
              - key: app.kubernetes.io/name
                operator: In
                values: ["simstudio"]
          topologyKey: topology.gke.io/zone

# Service Account with Workload Identity integration
serviceAccount:
  create: true
  annotations:
    iam.gke.io/gcp-service-account: "simstudio@your-project-id.iam.gserviceaccount.com"

# Additional environment variables for GCP service integration
extraEnvVars:
  - name: GOOGLE_APPLICATION_CREDENTIALS
    value: "/var/secrets/google/key.json"

# Additional volumes for service account credentials
extraVolumes:
  - name: google-cloud-key
    secret:
      secretName: google-service-account-key

extraVolumeMounts:
  - name: google-cloud-key
    mountPath: /var/secrets/google
    readOnly: true
# -----------------------------------------------------------------------------
# External Secrets Operator (ESO) — recommended for production on GCP
# -----------------------------------------------------------------------------
# To switch from inline secrets to ESO:
#   1. Install external-secrets in the cluster (https://external-secrets.io).
#   2. Create a ClusterSecretStore backed by Google Secret Manager and bind
#      Workload Identity from this namespace's ServiceAccount.
#   3. Remove entries from app.env above (or leave them and the chart will
#      fail-fast naming each unmapped key).
#   4. Uncomment the block below and replace the paths with the names of the
#      secrets in your project.
#
# externalSecrets:
#   enabled: true
#   secretStoreRef:
#     name: "gcp-secret-store"          # name of your ClusterSecretStore
#     kind: "ClusterSecretStore"
#   remoteRefs:
#     app:
#       BETTER_AUTH_SECRET:  "sim-better-auth-secret"
#       ENCRYPTION_KEY:      "sim-encryption-key"
#       INTERNAL_API_SECRET: "sim-internal-api-secret"
#       CRON_SECRET:         "sim-cron-secret"
#       API_ENCRYPTION_KEY:  "sim-api-encryption-key"
#       # Provider keys typical for a GCP deployment. Vertex AI auth is
#       # usually handled by Workload Identity, so VERTEX_CREDENTIALS is
#       # only needed when running outside the GCP credential chain.
#       OPENAI_API_KEY:      "sim-openai-api-key"
#       # ANTHROPIC_API_KEY: "sim-anthropic-api-key"
#       # VERTEX_CREDENTIALS: "sim-vertex-credentials-json"
#     postgresql:
#       password: "sim-postgresql-password"