How to pass userdata when creating an ec2-instance with the service catalog?

[zackproser]

A customer asked:

Could you please share the option to pass our userdata while creating an ec2-instance using the below the terragrunt module?
https://github.com/gruntwork-io/terraform-aws-service-catalog/tree/master/modules/services/ec2-instance which calls this module under the hood: https://github.com/gruntwork-io/terraform-aws-server/tree/v0.13.7/modules/single-server

[zackproser]

Let's first start by understanding how user data is currently configured and passed in the module.

In the ec2-instance service catalog module you indicated, the user-data script is specified in the user-data.sh file here (in the root of the same module).

You will notice that user-data.sh file is a template that has values expecting to be interpolated later, such as the following:

readonly users_for_ip_lockdown=(${ip_lockdown_users})
start_ec2_baseline \
  "${enable_cloudwatch_log_aggregation}" \
  "${enable_ssh_grunt}" \
  "${enable_fail2ban}" \
  "${enable_ip_lockdown}" \
  "${ssh_grunt_iam_group}" \
  "${ssh_grunt_iam_group_sudo}" \
  "${log_group_name}" \
  "${external_account_ssh_grunt_role_arn}" \
  "$${users_for_ip_lockdown[@]}"  # Need a double dollar-sign here to avoid Terraform interpolation


volume_json=$(echo ${ebs_volumes} | base64 -d)
for name in $(echo $${volume_json} | jq -r 'keys[]') ; do
    mount_point=$(echo $${volume_json} | jq -r ".\"$${name}\".mount_point")
    device_name=$(echo $${volume_json} | jq -r ".\"$${name}\".device_name")
    owner=$(echo $${volume_json} | jq -r ".\"$${name}\".owner")
    id=$(echo ${ebs_volume_data} | base64 -d | jq -r "[.\"$${name}\"][0].id")
    mount-ebs-volume \
        --aws-region "${ebs_aws_region}" \
        --volume-id "$${id}" \
        --device-name "$${device_name}" \
        --mount-point "$${mount_point}" \
        --owner "$${owner}"
done

On lines 97 to 114 of the main.tf file in that module, a local variable called base_user_data is created by making a call to the terraform function templatefile to load the above ^ user-data.sh template into memory after passing in the variables expected by the script.

With that done, a new local map is created in lines 77 to 81 here which represents the structure expected by the variable cloud_init_parts which is defined here on lines 154 to 162 of variables.tf. As noted in a comment there, this doc explains the use of template_cloudinit_config which is a generic definition for cloud-specific user data mechanisms such as AWS user-data.

All that said, here's the official guide to configuring user data within your own terraform / terragrunt config.