Terraform AWS Plan 與 Apply 工作流程

# Terraform AWS Infrastructure Workflow
# Plan, human review, apply, verify, update state
osop_version: "2.0"
id: terraform-aws
name: "Terraform AWS Plan 與 Apply 工作流程"

nodes:
  - id: tf_init
    type: infra
    purpose: Initialize Terraform working directory and backend
    runtime:
      tool: terraform
      action: init
      backend: s3
      backend_config:
        bucket: terraform-state-prod
        key: infra/terraform.tfstate
        region: us-east-1
    outputs: [init_status]
    timeout_sec: 120

  - id: tf_validate
    type: infra
    purpose: Validate Terraform configuration syntax and consistency
    runtime:
      tool: terraform
      action: validate
    inputs: [init_status]
    outputs: [validation_result]
    timeout_sec: 30

  - id: tf_plan
    type: infra
    purpose: Generate Terraform execution plan showing changes
    runtime:
      tool: terraform
      action: plan
      var_file: environments/production.tfvars
      out: tfplan.out
    inputs: [validation_result]
    outputs: [plan_output, resource_changes]
    timeout_sec: 300
    explain: |
      Creates a saved plan file showing all resources to be
      added, changed, or destroyed. The plan is saved for
      exact apply to prevent drift between plan and apply.

  - id: human_review
    type: human
    purpose: Infrastructure team reviews the Terraform plan
    role: platform-engineer
    inputs: [plan_output, resource_changes]
    approval_gate:
      required_approvers: 2
      timeout_min: 120
    explain: |
      Two platform engineers must approve the plan. They verify
      no unexpected resource deletions, cost impact is acceptable,
      and security groups/IAM changes are intentional.

  - id: tf_apply
    type: infra
    purpose: Apply the approved Terraform plan
    runtime:
      tool: terraform
      action: apply
      plan_file: tfplan.out
    inputs: [plan_output]
    outputs: [apply_result, resource_outputs]
    timeout_sec: 900

  - id: verify_infra
    type: api
    purpose: Verify deployed infrastructure is healthy
    runtime:
      endpoint: health-checks
      method: GET
      url: "https://{{resource_outputs.lb_dns}}/health"
    inputs: [resource_outputs]
    outputs: [health_status]
    retry_policy:
      max_retries: 5
      backoff_sec: 15
    timeout_sec: 120

  - id: notify_complete
    type: api
    purpose: Notify team of successful infrastructure update
    runtime:
      endpoint: slack-webhook
      method: POST
      url: "{{SLACK_WEBHOOK_URL}}"
    inputs: [apply_result, health_status]

edges:
  - from: tf_init
    to: tf_validate
    mode: sequential

  - from: tf_validate
    to: tf_plan
    mode: conditional
    condition: "validation_result.valid == true"

  - from: tf_plan
    to: human_review
    mode: sequential

  - from: human_review
    to: tf_apply
    mode: sequential

  - from: tf_apply
    to: verify_infra
    mode: sequential

  - from: verify_infra
    to: notify_complete
    mode: sequential