Kubernetes Infrastructure Patterns

Kubernetes Infrastructure Patterns

Infrastructure provisioning using Kubernetes operators, Helm, and GitOps practices.

Core Stack

Context7 Library IDs

Query these for current best practices:

• ArgoCD: /​argoproj/​argo-cd
• Helm: /​helm/​helm

Execution Rules

1. GitOps first. All changes through git, not kubectl apply ad-hoc
2. Helm best practices. Values.yaml for configuration, templates for logic
3. Security. No secrets in code, use External Secrets Operator
4. Idempotent. All operations safe to retry
5. Validate. Always helm template and kubectl diff before apply

Available Operators

Size Presets

Infrastructure Provisioning Process

Step 1: Parse Requirements

Extract infrastructure from task XML:

<infrastructure>
    <component type="postgresql" name="app-db">
        <size>small</​size>
        <replicas>1</​replicas>
        <database>app_production</​database>
    </​component>
</​infrastructure>

Step 2: Generate Manifests

Create manifests in the infra/ directory:

infra/
├── postgresql/
│   └── cluster.yaml
├── valkey/
│   └── redis.yaml
├── seaweedfs/
│   └── bucket-init.yaml
└── kustomization.yaml

Step 3: PostgreSQL Example

apiVersion: postgresql.cnpg.io/​v1
kind: Cluster
metadata:
  name: app-db
  namespace: databases
spec:
  instances: 1
  storage:
    size: 5Gi
    storageClass: mayastor
  bootstrap:
    initdb:
      database: app_production
      owner: app_user

Step 4: Valkey/​Redis Example

apiVersion: redis.redis.opstreelabs.in/​v1beta2
kind: Redis
metadata:
  name: app-cache
  namespace: databases
spec:
  kubernetesConfig:
    image: redis:7-alpine
  storage:
    volumeClaimTemplate:
      spec:
        storageClassName: mayastor
        accessModes: ["ReadWriteOnce"]
        resources:
          requests:
            storage: 1Gi

Step 5: Apply and Wait

# Apply manifests
kubectl apply -k infra/

# Wait for PostgreSQL
kubectl wait --for=condition=Ready cluster/​app-db -n databases --timeout=300s

# Wait for Valkey
kubectl wait --for=condition=Ready redis/​app-cache -n databases --timeout=300s

Step 6: Create Infrastructure ConfigMap

apiVersion: v1
kind: ConfigMap
metadata:
  name: app-infra-config
  namespace: app
  labels:
    cto.platform/​type: infrastructure-config
data:
  DATABASE_URL: postgresql://app_user:[email protected]:5432/​app
  DATABASE_HOST: app-db-rw.databases.svc
  DATABASE_PORT: "5432"
  DATABASE_NAME: app
  
  REDIS_URL: redis://app-cache.databases.svc:6379
  REDIS_HOST: app-cache.databases.svc
  REDIS_PORT: "6379"
  
  S3_ENDPOINT: http://seaweedfs-filer.seaweedfs.svc:8333
  S3_BUCKET: app-uploads

Validation Commands

# Helm validation
helm lint ./​chart
helm template ./​chart --debug

# Kubernetes validation
kubectl diff -f manifest.yaml
kubeval manifest.yaml

# ArgoCD
argocd app diff app-name

# Check status
kubectl get all -n databases
kubectl get cluster -n databases -o wide
kubectl get redis -n databases

Error Handling

If provisioning fails:

1. Check operator logs: kubectl logs -n operators -l app.kubernetes.io/​name=<operator>
2. Describe the resource: kubectl describe cluster/​app-db -n databases
3. Check events: kubectl get events -n databases --sort-by='.lastTimestamp'
4. Verify storage class: kubectl get storageclass mayastor

Guidelines

• Use operators for stateful services (databases, caches)
• Store connection details in ConfigMaps for other agents
• Always wait for resources to be ready before completing
• Document connection information in infra/​README.md
• Use GitOps (ArgoCD) for production deployments
• Never hardcode secrets in manifests