K8s Deployment Tools & Terraform

---

Terraform Is Not a K8s Deployment Tool

Terraform is an infrastructure management tool. K8s is just one of many things it can manage.

What Terraform Can Manage

AWS       → VPC, EKS, EC2, RDS, S3, IAM, Route53, ALB...
Azure     → AKS, VM, Storage, AD...
GCP       → GKE, Compute, Cloud SQL...
K8s       → Helm release, kubectl manifest, namespace...
Others    → GitHub repo, Datadog monitor, PagerDuty,
            Cloudflare DNS, Vault secrets...

Basically anything with an API can be managed by Terraform.

---

K8s Deployment Tools Comparison

Tool	Purpose	Best For
Terraform + Helm	Manage infra + deploy apps together	Ops/Platform teams
Helm standalone	K8s app deployment	K8s only, no cloud resources
ArgoCD / FluxCD	GitOps — git push auto-deploys to K8s	Continuous deployment, app teams
kubectl	Manual YAML apply	Debugging, learning
Kustomize	YAML overlay for multi-env	When you don't want Helm templates
Pulumi	Write infra in Python/TypeScript	When you don't want to learn HCL

---

Terraform + Helm Together

Terraform manages:
  → K8s cluster itself (EKS / AKS / GKE)
  → Cloud resources (Load Balancer, IAM, Secrets, Storage...)
  → Calls Helm to deploy apps

Helm manages:
  → Apps inside K8s (Deployment, Service, ConfigMap, RBAC...)

If you only use Helm — you can deploy apps to K8s, but who creates the K8s cluster? Load Balancer? Secrets?

Terraform manages the full stack (cloud + K8s), Helm only manages apps inside K8s.

---

Helm Chart vs Manual YAML

Helm chart — pre-built YAML for you

Manual:
  Hand-write Deployment YAML
  Hand-write Service YAML
  Hand-write ServiceAccount + RBAC YAML
  kubectl apply -f one by one

Helm:
  Chart already contains all YAML templates
  Just fill in values (image tag, replica count...)
  helm install does everything at once

kubectl_manifest — still write YAML yourself

# Inside Terraform, but the YAML is yours
yaml_body = <<-YAML
  apiVersion: apps/v1
  kind: Deployment
  ...
YAML

Comparison

	Manual kubectl	Terraform + Helm
Complex components	Write 20+ YAML files yourself	Helm chart pre-built, fill values
Simple resources	kubectl apply -f	kubectl_manifest (still write YAML)
Version control	Manage YAML files manually	git + Terraform state auto-tracking
Multi-env	Copy YAML per env and modify	Same code, different inputs
Rollback	Apply old YAML	Terraform state tracks, auto diff

Terraform's value is not "writing YAML for you" — it's state management + multi-env + version control + dependency ordering.

---

How Helm set {} Becomes YAML

Helm's set uses . and [0] to represent YAML hierarchy. Terraform's set blocks are assembled by Helm into complete K8s YAML.

Example: extraSecretMounts

What you write in Terraform:

set {
  name  = "server.extraSecretMounts[0].name"
  value = "my-secret"
}
set {
  name  = "server.extraSecretMounts[0].secretName"
  value = "k8s-secret-name"
}
set {
  name  = "server.extraSecretMounts[0].mountPath"
  value = "/etc/secrets/my-secret"
}
set {
  name  = "server.extraSecretMounts[0].readOnly"
  value = "true"
}

What Helm generates:

volumes:
  - name: my-secret
    secret:
      secretName: k8s-secret-name

containers:
  - name: server
    volumeMounts:
      - name: my-secret
        mountPath: /etc/secrets/my-secret
        readOnly: true

Reading set name field

server.extraSecretMounts[0].secretName
  │       │                │     │
  │       │                │     └── YAML key
  │       │                └── First element of array
  │       └── Key in Helm values
  └── This setting belongs to server section

values = [file(...)] vs set {}

resource "helm_release" "app" {
  # Method 1: YAML file for bulk settings
  values = [file("${path.module}/config.yaml")]

  # Method 2: set for individual overrides (higher priority)
  set {
    name  = "server.replicaCount"
    value = "2"
  }
}

Both can be used together. set takes priority over values file.