Terraform 基礎架構部署

# Terraform Infrastructure Deployment
# Plan, review, apply infrastructure changes with drift detection and notifications
osop_version: "2.0"
id: terraform-deploy
name: "Terraform 基礎架構部署"

nodes:
  - id: init_workspace
    type: cli
    purpose: Initialize Terraform workspace and download provider plugins
    runtime:
      command: "terraform init -backend-config=env/${ENVIRONMENT}.hcl -reconfigure"
    outputs: [workspace_ready]
    timeout_sec: 120

  - id: validate_config
    type: cli
    purpose: Validate Terraform configuration syntax and internal consistency
    runtime:
      command: "terraform validate && terraform fmt -check -recursive"
    inputs: [workspace_ready]
    outputs: [validation_result]
    timeout_sec: 60

  - id: plan
    type: cli
    purpose: Generate and save an execution plan showing proposed infrastructure changes
    runtime:
      command: "terraform plan -out=tfplan -detailed-exitcode -var-file=env/${ENVIRONMENT}.tfvars"
    inputs: [validation_result]
    outputs: [plan_file, change_summary]
    timeout_sec: 300
    explain: |
      Exit code 2 means changes detected. The plan file is saved as an artifact
      for exact reproducibility during the apply phase.

  - id: security_scan
    type: cli
    purpose: Scan the plan for security misconfigurations using tfsec and checkov
    runtime:
      command: "tfsec . --format json && checkov -d . --framework terraform --output json"
    inputs: [plan_file]
    outputs: [security_findings]
    timeout_sec: 180

  - id: cost_estimate
    type: api
    purpose: Estimate monthly cost impact of planned changes via Infracost
    runtime:
      endpoint: /api/v1/estimate
      method: POST
      url: https://infracost.internal
    inputs: [plan_file]
    outputs: [cost_delta, monthly_estimate]
    security:
      auth: bearer_token
      secret_ref: INFRACOST_API_KEY
    timeout_sec: 60

  - id: approval_gate
    type: human
    purpose: Platform engineer reviews plan, security findings, and cost estimate before apply
    role: platform_engineer
    inputs: [change_summary, security_findings, cost_delta]
    approval_gate:
      required_approvers: 1
      timeout_min: 120
    explain: |
      Approver must verify no destructive changes to stateful resources,
      security findings are acceptable, and cost impact is within budget.

  - id: apply
    type: cli
    purpose: Apply the saved plan to provision or update infrastructure
    runtime:
      command: "terraform apply -auto-approve tfplan"
    inputs: [plan_file]
    outputs: [apply_result, resource_ids]
    timeout_sec: 600
    retry_policy:
      max_retries: 1
      backoff_sec: 30
    security:
      credentials: [AWS_ACCESS_KEY_ID, AWS_SECRET_ACCESS_KEY]

  - id: verify_state
    type: cli
    purpose: Run post-apply drift detection and verify expected resources exist
    runtime:
      command: "terraform plan -detailed-exitcode && ./scripts/verify-resources.sh"
    inputs: [apply_result]
    outputs: [drift_status]
    timeout_sec: 120

  - id: notify_team
    type: api
    purpose: Post deployment summary to Slack with change details and cost impact
    runtime:
      endpoint: /api/chat.postMessage
      method: POST
      url: https://slack.com
    inputs: [apply_result, cost_delta, drift_status]
    security:
      auth: bearer_token
      secret_ref: SLACK_BOT_TOKEN

edges:
  - from: init_workspace
    to: validate_config
    mode: sequential

  - from: validate_config
    to: plan
    mode: sequential

  - from: plan
    to: security_scan
    mode: parallel

  - from: plan
    to: cost_estimate
    mode: parallel

  - from: security_scan
    to: approval_gate
    mode: sequential

  - from: cost_estimate
    to: approval_gate
    mode: sequential

  - from: approval_gate
    to: apply
    mode: sequential

  - from: apply
    to: verify_state
    mode: sequential

  - from: verify_state
    to: notify_team
    mode: sequential

  - from: verify_state
    to: plan
    mode: fallback
    label: "Drift detected, re-plan"