A Cloud Function that automatically recovers terminated spot VMs in Google Cloud Platform.

• Periodically checks spot VMs
• Automatically restarts terminated VMs
• Scheduled execution with Cloud Scheduler
• VM exclusion with labels:

  # Exclude a VM from recovery
  gcloud compute instances add-labels INSTANCE_NAME --labels=exclude_from_keeper=true
  
  # Include a VM back
  gcloud compute instances remove-labels INSTANCE_NAME --labels=exclude_from_keeper

• Terraform >= 1.0
• Google Cloud Account
• Google Cloud CLI installed and configured
• Required IAM permissions:
  • Cloud Functions Admin
  • Cloud Scheduler Admin
  • Storage Admin
  • Service Account Admin
  • IAM Admin

keeper/
├── main.py              # Cloud Function source code
├── requirements.txt     # Python dependencies
├── terraform/          # Infrastructure as Code
│   ├── main.tf         # Main Terraform configuration
│   ├── variables.tf    # Variable definitions
│   └── terraform.tfvars.example  # Example variable values

1. Clone the repository
2. Authenticate with Google Cloud:

   gcloud auth application-default login

3. Create terraform.tfvars:

   cd keeper/terraform
   cp terraform.tfvars.example terraform.tfvars

4. Edit terraform.tfvars:

   project_id = "your-project-id"
   region     = "your-region"
   zone       = "your-zone"
   schedule   = "*/1 * * * *"  # Desired cron schedule

5. Initialize Terraform:

   terraform init

6. Review the plan:

   terraform plan

7. Apply the configuration:

   terraform apply

You can configure the following variables in terraform.tfvars:

• project_id: Your GCP project ID
• region: Region where resources will be created
• zone: Zone where spot VMs are located
• schedule: Cloud Scheduler cron expression

To remove all created resources:

terraform destroy