Cortex Resource Manager

""" Example usage of Worker Management tools for Resource Manager MCP Server This file demonstrates how to use the worker management functionality. """ from src.worker_manager import WorkerManager, WorkerManagerError def main(): """Demonstrate worker management functionality""" # Initialize the worker manager # In production, you would load configuration from a file config = { "talos_mcp_endpoint": "stdio://talos-mcp-server", "proxmox_mcp_endpoint": "stdio://proxmox-mcp-server", "kubectl_context": None # Use default context } manager = WorkerManager(config) print("=" * 80) print("Worker Management Examples") print("=" * 80) # Example 1: List all workers print("\n1. Listing all workers:") print("-" * 80) try: all_workers = manager.list_workers() print(f"Total workers: {len(all_workers)}") for worker in all_workers: print(f"\nWorker: {worker['name']}") print(f" Status: {worker['status']}") print(f" Type: {worker['type']}") resources = worker['resources']['allocatable'] print(f" Resources:") print(f" CPU: {resources.get('cpu', 'unknown')}") print(f" Memory: {resources.get('memory', 'unknown')}") if worker['type'] == 'burst' and 'ttl_expires' in worker: print(f" TTL Expires: {worker['ttl_expires']}") except WorkerManagerError as e: print(f"Error listing workers: {e}") # Example 2: List only burst workers print("\n\n2. Listing burst workers only:") print("-" * 80) try: burst_workers = manager.list_workers(type_filter="burst") print(f"Burst workers: {len(burst_workers)}") for worker in burst_workers: print(f"\nBurst Worker: {worker['name']}") print(f" Status: {worker['status']}") if 'ttl_expires' in worker: print(f" TTL Expires: {worker['ttl_expires']}") except WorkerManagerError as e: print(f"Error listing burst workers: {e}") # Example 3: List only permanent workers print("\n\n3. Listing permanent workers only:") print("-" * 80) try: permanent_workers = manager.list_workers(type_filter="permanent") print(f"Permanent workers: {len(permanent_workers)}") for worker in permanent_workers: print(f"\nPermanent Worker: {worker['name']}") print(f" Status: {worker['status']}") except WorkerManagerError as e: print(f"Error listing permanent workers: {e}") # Example 4: Provision burst workers print("\n\n4. Provisioning burst workers:") print("-" * 80) print("NOTE: This is a dry-run example. Actual provisioning requires Talos/Proxmox MCP.") try: # Uncomment to actually provision workers # workers = manager.provision_workers(count=2, ttl=24, size="medium") # Simulated output print("Would provision 2 medium burst workers with 24-hour TTL:") print(" - burst-worker-1234567890-0") print(" Size: medium (4 CPU, 8GB RAM, 100GB disk)") print(" TTL: 24 hours") print(" - burst-worker-1234567890-1") print(" Size: medium (4 CPU, 8GB RAM, 100GB disk)") print(" TTL: 24 hours") except WorkerManagerError as e: print(f"Error provisioning workers: {e}") # Example 5: Get worker details print("\n\n5. Getting worker details:") print("-" * 80) try: # Get details for the first worker all_workers = manager.list_workers() if all_workers: worker_id = all_workers[0]['name'] details = manager.get_worker_details(worker_id) print(f"Worker: {details['name']}") print(f"Type: {details['type']}") print(f"Status: {details['status']}") print(f"Created: {details['created']}") print("\nResources:") capacity = details['resources']['capacity'] allocatable = details['resources']['allocatable'] print(f" Capacity:") print(f" CPU: {capacity.get('cpu', 'unknown')}") print(f" Memory: {capacity.get('memory', 'unknown')}") print(f" Pods: {capacity.get('pods', 'unknown')}") print(f" Allocatable:") print(f" CPU: {allocatable.get('cpu', 'unknown')}") print(f" Memory: {allocatable.get('memory', 'unknown')}") print(f" Pods: {allocatable.get('pods', 'unknown')}") print("\nConditions:") for condition in details['conditions']: print(f" {condition.get('type')}: {condition.get('status')}") except WorkerManagerError as e: print(f"Error getting worker details: {e}") # Example 6: Drain a worker (commented out for safety) print("\n\n6. Draining a worker:") print("-" * 80) print("NOTE: This is a dry-run example. Uncomment to actually drain a worker.") # Uncomment to actually drain a worker # try: # worker_id = "burst-worker-1234567890-0" # result = manager.drain_worker(worker_id) # print(f"Worker: {result['worker_id']}") # print(f"Status: {result['status']}") # print(f"Message: {result['message']}") # except WorkerManagerError as e: # print(f"Error draining worker: {e}") print("Would drain worker by:") print(" 1. Marking node as unschedulable") print(" 2. Evicting all pods (except DaemonSets)") print(" 3. Waiting for graceful pod termination (5 min grace period)") # Example 7: Destroy a burst worker (commented out for safety) print("\n\n7. Destroying a burst worker:") print("-" * 80) print("NOTE: This is a dry-run example. Uncomment to actually destroy a worker.") print("IMPORTANT: Always drain workers before destroying!") # Safe workflow example (commented out) # try: # worker_id = "burst-worker-1234567890-0" # # # Step 1: Verify it's a burst worker # details = manager.get_worker_details(worker_id) # if details['type'] != 'burst': # print(f"SAFETY VIOLATION: Cannot destroy permanent worker {worker_id}") # else: # # Step 2: Drain the worker # drain_result = manager.drain_worker(worker_id) # print(f"Drained: {drain_result['status']}") # # # Step 3: Destroy the worker # destroy_result = manager.destroy_worker(worker_id) # print(f"Destroyed: {destroy_result['status']}") # print(f"Cluster removal: {destroy_result['removed_from_cluster']}") # print(f"VM deletion: {destroy_result['vm_deleted']}") # # except WorkerManagerError as e: # print(f"Error destroying worker: {e}") print("Safe destroy workflow:") print(" 1. Verify worker is a burst worker (not permanent)") print(" 2. Drain the worker to move pods") print(" 3. Remove worker from Kubernetes cluster") print(" 4. Delete the VM via Talos/Proxmox MCP") print("\nSAFETY FEATURES:") print(" - Only burst workers can be destroyed") print(" - Permanent workers are protected") print(" - Requires drain before destroy (unless force=True)") # Example 8: Error handling print("\n\n8. Error handling examples:") print("-" * 80) # Try to destroy a permanent worker (will fail) print("\nAttempting to destroy a permanent worker (will fail):") try: # This would fail with a safety violation # manager.destroy_worker("permanent-worker-1", force=True) print("BLOCKED: WorkerManagerError - SAFETY VIOLATION") print("Cannot destroy permanent worker. Only burst workers can be destroyed.") except WorkerManagerError as e: print(f"Error (expected): {e}") # Try to get details for non-existent worker print("\nAttempting to get details for non-existent worker:") try: manager.get_worker_details("non-existent-worker") except WorkerManagerError as e: print(f"Error (expected): {e}") print("\n" + "=" * 80) print("Examples complete!") print("=" * 80) if __name__ == "__main__": main()