Temporal Terraform Orchestrator

Problem

Managing Terraform deployments across multiple workspaces with interdependencies is challenging:

• Manual dependency tracking: Teams must manually determine which workspaces depend on others and execute them in the correct order.
• Output propagation: Passing outputs from one workspace as inputs to another requires manual intervention or brittle scripting.
• No parallelization: Without proper orchestration, independent workspaces run sequentially, wasting time.
• Lack of durability: Failed deployments require manual recovery and re-execution.

Approach

I built a Temporal-based orchestration system that treats multi-workspace Terraform deployments as durable workflows with automatic dependency resolution.

Architecture

The system consists of two main workflow types:

• ParentWorkflow: Orchestrates the overall deployment by building a directed acyclic graph (DAG) of workspace dependencies, validating configuration, and managing completion signals.
• TerraformWorkflow: Executes individual workspace operations (init, validate, plan, apply) and captures outputs for downstream consumption.

Key Features

• Automatic dependency resolution Workspaces declare their dependencies, and the system builds a DAG to determine execution order. Independent workspaces run concurrently for faster deployments.

• Output propagation Terraform outputs from completed workspaces automatically map to input variables in downstream workspaces through configurable input mappings.

• Temporal durability Workflows survive worker restarts and failures. Temporal handles automatic retries with exponential backoff, making deployments resilient.

• MCP server integration AI agents can trigger deployments and monitor execution status through standardized Model Context Protocol tools:

  • list_workflows: Returns configured workspaces and relationships
  • execute_workflow: Initiates deployment with provided configuration
  • get_workflow_status: Checks execution status by workflow ID

Outcomes

• Faster deployments: Independent workspaces run in parallel, reducing overall deployment time.
• Reliable infrastructure changes: Temporal’s durability guarantees mean deployments complete even through failures.
• AI-assisted operations: MCP integration enables AI agents to manage infrastructure deployments conversationally.
• Clear dependency visualization: The DAG representation makes workspace relationships explicit and auditable.

Key Contributions

• Designed and implemented a Temporal workflow architecture for multi-workspace Terraform orchestration.
• Built automatic dependency resolution using DAG construction with cycle detection.
• Implemented output-to-input mapping for seamless variable propagation between workspaces.
• Created an MCP server enabling AI-driven infrastructure management.
• Documented deployment patterns and configuration schemas for production use.