Resource Graphs

Resource graphs are the fundamental building blocks of TypeKro. They define collections of related Kubernetes resources with type-safe schemas and runtime dependencies. This guide covers everything you need to know about creating, composing, and deploying resource graphs.

What is a Resource Graph?

A resource graph is a typed collection of Kubernetes resources that can reference each other's properties at runtime. Think of it as a blueprint for infrastructure that includes:

Resource definitions - The actual Kubernetes resources (Deployments, Services, etc.)
Cross-resource references - Dynamic connections between resources
Type-safe schema - Input/output contracts with validation
Status mapping - How runtime state is exposed to users

typescript

import { type } from 'arktype';
import { kubernetesComposition, Cel } from 'typekro'; import { Deployment, Service, Ingress } from 'typekro/simple';

const WebAppSpec = type({
  name: 'string',
  image: 'string',
  replicas: 'number'
});

const WebAppStatus = type({
  url: 'string',
  phase: 'string'
});

const webApp = kubernetesComposition({
  // Graph definition
  {
    name: 'webapp',
    apiVersion: 'example.com/v1',
    kind: 'WebApp',
    spec: WebAppSpec,
    status: WebAppStatus
  },
  
  // Resource builder - defines what resources to create
  (schema) => ({
    deployment: Deployment({
      name: schema.spec.name,
      image: schema.spec.image,
      replicas: schema.spec.replicas
    }),
    service: Service({
      name: Cel.template('%s-service', schema.spec.name),
      selector: { app: schema.spec.name },
      ports: [{ port: 80, targetPort: 3000 }]
    })
  }),
  
  // Status builder - defines how to compute status
  (schema, resources) => ({
    url: Cel.template('http://%s', resources.service.status.loadBalancer.ingress[0].ip),
    phase: resources.deployment.status.phase
  })
);

Resource Graph Components

1. Graph Definition

The first parameter defines the resource graph metadata:

typescript

const definition = {
  name: 'my-application',           // Must be a valid Kubernetes resource name
  apiVersion: 'example.com/v1',     // Kubernetes-style API version
  kind: 'MyApplication',            // Resource type name
  spec: MyAppSpec,                  // Input schema (required)
  status: MyAppStatus               // Output schema (optional)
};

2. Resource Builder

The resource builder function creates the actual Kubernetes resources:

typescript

const resourceBuilder = (schema) => ({
  // Each key becomes a resource ID
  database: Deployment({
    name: Cel.template('%s-db', schema.spec.name),
    image: 'postgres:15',
    env: {
      POSTGRES_DB: schema.spec.database.name,
      POSTGRES_USER: schema.spec.database.user,
      POSTGRES_PASSWORD: schema.spec.database.password
    },
    ports: [{ containerPort: 5432 }]
  }),
  
  app: Deployment({
    name: schema.spec.name,
    image: schema.spec.image,
    env: {
      // Cross-resource reference
      DATABASE_HOST: 'postgres-service' // Use service name for DNS resolution
    }
  }),
  
  service: Service({
    name: Cel.template('%s-service', schema.spec.name),
    selector: { app: schema.spec.name },
    ports: [{ port: 80, targetPort: 3000 }]
  })
});

3. Status Builder

The status builder computes dynamic status based on resource state:

typescript

const statusBuilder = (schema, resources) => ({
  // Simple field mapping
  phase: resources.app.status.phase,
  readyReplicas: resources.app.status.readyReplicas,
  
  // Computed values
  url: Cel.expr<string>(
    schema.spec.environment, ' == "production" ? ',
    Cel.template('https://%s', resources.service.status.loadBalancer.ingress[0].hostname),
    ' : ',
    Cel.template('http://%s', resources.service.spec.clusterIP)
  ),
    
  // Complex logic with CEL expressions
  ready: Cel.expr<boolean>(
    resources.app.status.readyReplicas, 
    ' > 0 && ',
    resources.database.status.readyReplicas,
    ' > 0'
  )
});

Resource Graph Patterns

Basic Application Stack

typescript

const basicStack = kubernetesComposition({
  {
    name: 'basic-stack',
    apiVersion: 'example.com/v1',
    kind: 'BasicStack',
    spec: BasicStackSpec,
    status: type({ ready: 'boolean' })
  },
  (schema) => ({
    app: Deployment({
      name: schema.spec.name,
      image: schema.spec.image,
      replicas: schema.spec.replicas
    }),
    
    service: Service({
      name: Cel.template('%s-service', schema.spec.name),
      selector: { app: schema.spec.name },
      ports: [{ port: 80, targetPort: 3000 }]
    })
  }),
  (schema, resources) => ({
    ready: Cel.expr<boolean>(resources.app.status.readyReplicas, ' > 0')
  })
);

Resource Graph Validation

Schema Validation

typescript

import { type } from 'arktype';

const StrictAppSpec = type({
  name: 'string>2',  // At least 3 characters
  image: 'string',
  replicas: 'number>0',  // At least 1
  environment: '"development" | "staging" | "production"',
  resources: {
    cpu: 'string',
    memory: 'string'
  }
});

// Validation happens automatically
const validatedApp = kubernetesComposition({
  {
    name: 'validated-app',
    apiVersion: 'example.com/v1',
    kind: 'ValidatedApp',
    spec: StrictAppSpec,
    status: type({ ready: 'boolean' })
  },
  resourceBuilder,
  statusBuilder
);

Runtime Validation

typescript

try {
  const factory = graph.factory('direct');
  await factory.deploy(spec);
} catch (error) {
  if (error instanceof SchemaValidationError) {
    console.error('Invalid spec:', error.errors);
  }
}

Deployment Strategies

Direct Deployment

typescript

const factory = graph.factory('direct', {
  namespace: 'development',
  timeout: 300000
});

const instance = await factory.deploy({
  name: 'my-app',
  image: 'nginx:latest',
  replicas: 3
});

YAML Generation

typescript

// Generate ResourceGraphDefinition
const rgdYaml = graph.toYaml();

// Generate instance YAML
// Generate instance YAML using factory
const factory = graph.factory('kro');
const instanceYaml = factory.toYaml({
  name: 'my-app',
  image: 'nginx:latest',
  replicas: 3
});

KRO Integration

typescript

const kroFactory = graph.factory('kro', {
  namespace: 'production'
});

await kroFactory.deploy(spec);

Best Practices

1. Resource Naming

typescript

// ✅ Use consistent naming patterns
const resources = {
  database: Deployment({ name: Cel.template('%s-db', schema.spec.name) }),
  databaseService: Service({ name: Cel.template('%s-db-service', schema.spec.name) }),
  app: Deployment({ name: schema.spec.name }),
  appService: Service({ name: Cel.template('%s-service', schema.spec.name) })
};

// ❌ Avoid inconsistent naming
const badResources = {
  db: Deployment({ name: 'database' }),
  dbSvc: Service({ name: 'db-service' }),
  application: Deployment({ name: 'app' }),
  service: Service({ name: 'svc' })
};

2. Environment Configuration

typescript

// ✅ Use environment-specific logic
const getConfig = (env: string) => ({
  development: { replicas: 1, resources: { cpu: '100m', memory: '256Mi' } },
  staging: { replicas: 2, resources: { cpu: '200m', memory: '512Mi' } },
  production: { replicas: 5, resources: { cpu: '500m', memory: '1Gi' } }
}[env] || { replicas: 1, resources: { cpu: '100m', memory: '256Mi' } });

3. Resource Organization

typescript

// ✅ Group related resources logically
const resources = {
  // Database tier
  database: Deployment({ /* ... */ }),
  databaseService: Service({ /* ... */ }),
  
  // Application tier
  app: Deployment({ /* ... */ }),
  appService: Service({ /* ... */ }),
  
  // Configuration
  config: ConfigMap({ /* ... */ }),
  secrets: Secret({ /* ... */ })
};

4. Status Design

typescript

// ✅ Provide meaningful status
const statusBuilder = (schema, resources) => ({
  // Deployment state
  phase: resources.app.status.phase,
  readyReplicas: resources.app.status.readyReplicas,
  totalReplicas: resources.app.spec.replicas,
  
  // Service information
  serviceEndpoint: Cel.template('http://%s:80', resources.service.spec.clusterIP),
  
  // Overall health
  healthy: Cel.expr<boolean>(
    resources.app.status.readyReplicas, 
    ' == ', 
    resources.app.spec.replicas
  ),
  
  // Timestamp
  lastUpdated: new Date().toISOString()
});

Troubleshooting

Common Issues

Schema validation errors:

typescript

// Check your type definitions
const AppSpec = type({
  name: 'string',
  replicas: 'number'  // Make sure types match usage
});

Resource reference errors:

typescript

// Ensure referenced resources exist in the same graph
const resources = {
  database: Deployment({ name: 'db' }),
  app: Deployment({
    env: {
      DB_HOST: database.status.podIP  // 'database' must be defined above
    }
  })
};

Circular dependencies:

typescript

// ❌ Avoid circular references
const serviceA = Service({
  selector: { app: serviceB.metadata.labels.app }  // References B
});
const serviceB = Service({
  selector: { app: serviceA.metadata.labels.app }  // References A
});

Next Steps

Status Hydration - Learn how status is computed and updated
Cross-Resource References - Deep dive into resource interconnection
CEL Expressions - Add runtime logic to your graphs
Direct Deployment - Deploy graphs directly to Kubernetes

Resource Graphs ​

What is a Resource Graph? ​

Resource Graph Components ​

1. Graph Definition ​

2. Resource Builder ​

3. Status Builder ​

Resource Graph Patterns ​

Basic Application Stack ​

Resource Graph Validation ​

Schema Validation ​

Runtime Validation ​

Deployment Strategies ​

Direct Deployment ​

YAML Generation ​

KRO Integration ​

Best Practices ​

1. Resource Naming ​

2. Environment Configuration ​

3. Resource Organization ​

4. Status Design ​

Troubleshooting ​

Common Issues ​

Next Steps ​

Resource Graphs

What is a Resource Graph?

Resource Graph Components

1. Graph Definition

2. Resource Builder

3. Status Builder

Resource Graph Patterns

Basic Application Stack

Resource Graph Validation

Schema Validation

Runtime Validation

Deployment Strategies

Direct Deployment

YAML Generation

KRO Integration

Best Practices

1. Resource Naming

2. Environment Configuration

3. Resource Organization

4. Status Design

Troubleshooting

Common Issues

Next Steps