Skip to main content

Troubleshooting

Common Issues​

Terraform State Locked​

cd iac/aws
terraform force-unlock <LOCK_ID>
# Or use the script
./scripts/setup_classroom.sh --name my-class --cloud aws --force-unlock

Lambda Packaging Fails​

# Install missing dependencies
pip3 install virtualenv
./scripts/package_lambda.sh --cloud aws

AWS Credentials Not Found​

aws configure
# Or use environment variables
export AWS_ACCESS_KEY_ID="your_key"
export AWS_SECRET_ACCESS_KEY="your_secret"

Backend Already Exists​

  • The script handles existing backends gracefully
  • Use --destroy to clean up if needed

Deployment Issues​

Resources Already Exist​

If Terraform tries to create resources that already exist:

  1. Check if resources are in Terraform state:

    cd iac/aws
    terraform state list

  2. If resources exist in AWS but not in state:

    • This shouldn't happen with clean deployments
    • Use --destroy then redeploy for a fresh start

  3. If resources exist in state but Terraform wants to recreate:

    • Run terraform refresh to sync state
    • Then terraform plan to see what Terraform wants to do

Deployment Fails Midway​

If deployment fails:

  1. Check the error message - It usually tells you what went wrong
  2. Check Terraform state: 
    cd iac/aws
    terraform state list
  3. Retry the deployment - Terraform is idempotent and will continue where it left off
  4. If stuck, use Clean Slate - Destroy and recreate

Frontend Not Updating​

If frontend changes aren't showing:

  1. Rebuild and redeploy frontend:

    ./scripts/build_frontend.sh --environment dev --region eu-west-3

  2. Clear CloudFront cache (if using custom domain):

    • Go to AWS Console → CloudFront
    • Select your distribution
    • Click "Invalidations" → "Create invalidation"
    • Enter /* and create

CloudFront Issues​

Certificate Validation Fails​

  • Verify DNS records are correctly added in GoDaddy
  • Check for typos in the CNAME record values
  • Wait longer (DNS can take up to 48 hours, but usually 5-10 minutes)
  • Verify the certificate status in AWS Console (ACM in us-east-1)

CloudFront Distribution Not Created​

  • Verify wait_for_certificate_validation = true in main.tf
  • Check that certificate status is ISSUED (not PENDING_VALIDATION)
  • Run terraform plan to see what Terraform wants to create
  • Check for errors in terraform apply output

DNS Not Resolving​

  • Verify CNAME records are added correctly in GoDaddy
  • Check DNS propagation: dig ec2-management.testingfantasy.com
  • Wait longer for DNS propagation (can take up to 48 hours)
  • Verify CloudFront distribution is Deployed (not In Progress)

403 AccessDeniedException from CloudFront​

  • This is usually a CloudFront configuration issue
  • Verify the origin (Lambda Function URL) is correct
  • Check CloudFront logs in CloudWatch
  • Verify the Lambda function URL is publicly accessible

CNAMEAlreadyExists Error​

This happens when:

  • An old CloudFront distribution was deleted, but its CNAME record still exists in GoDaddy
  • Another CloudFront distribution already uses the same domain

Solution:

  1. Remove the old CNAME record from GoDaddy
  2. Set wait_for_certificate_validation = false temporarily
  3. Run terraform apply to create the new distribution
  4. Add the new CNAME record pointing to the new CloudFront domain

Debug Mode​

Enable verbose Terraform output:

export TF_LOG=DEBUG
./scripts/setup_classroom.sh --name debug-class --cloud aws