Terraform module to create an AWS AMI snapshot-image that can subsequently be used to start an OPNsense instance within AWS.
Switch branches/tags
Nothing to show
Clone or download
Fetching latest commit…
Cannot retrieve the latest commit at this time.
Permalink
Type Name Latest commit message Commit time
Failed to load latest commit information.
data
.gitignore
LICENSE
README.md
main.tf
outputs.tf
variables.tf

README.md

Image Create :: OPNsense on AWS

Terraform module to create an AWS AMI snapshot-image that can subsequently be used to start an OPNsense instance within AWS.

This module provides an boot-time syshook for OPNsense that collects input parameters from the AWS meta-data service at http://169.254.169.254 and applies it as required to the OPNsense config.xml file at boot. Config attributes that are managed this way include:-

  • root user sshkey
  • Public and Private network interface cards
  • IPv4 address, subnet, gateway, dns
  • IPv6 address, subnet, gateway

This allows the resulting OPNsense AMI to be used in regular Terraform devops automation situations. Additionally, users of the resulting OPNsense instance can inject user-data scripts at initial instance boot, however the system does not include cloudinit, so user-data scripts need to be contained /bin/sh scripts.

Usage

This module is mildly unusual in that the final result does not provide a running EC2 Instance. The correct behaviour of this module will result in an AWS AMI while the EV2 instance used in the process of creating the image will poweroff.

The example below shows an example setup - note that the root_passwd variable is optional and by default will use the same default password that OPNsense uses, that is "opnsense" - be smart, change this because your OPNsense instance will be publicly accessible to begin with unless you take other measures to prevent inbound access to TCP22 (SSH) and TCP443 (HTTPS).

variable "aws_access_key_id" {}       # set via environment value `TF_VAR_aws_access_key_id`
variable "aws_secret_access_key" {}   # set via environment value `TF_VAR_aws_secret_access_key`

module "opnsense-image" {
  source  = "verbnetworks/opnsense-image/aws"

  opnsense_release = "18.7"
  root_passwd = "honeyPot..."

  aws_region = "ap-southeast-1"
  aws_access_key_id = "${var.aws_access_key_id}"
  aws_secret_access_key = "${var.aws_secret_access_key}"

  do_opnsense_install = 1
  do_cleanup_shutdown = 1
  do_image = 1
  do_self_destruct = 1
}

output "provider" { value = "${module.opnsense-image.provider}"}
output "region" { value = "${module.opnsense-image.region}"}
output "build_id" { value = "${module.opnsense-image.build_id}"}
output "image_name" { value = "${module.opnsense-image.image_name}"}
output "image_action_outfile" { value = "${module.opnsense-image.image_action_outfile}"}

After the build process completes you should observe among the final Terraform log lines the following, thus indicating the build process is complete and the image is taking place on the Digital Ocean backend.

action-status (local-exec): !!!!
action-status (local-exec): !!!! build_id: YDYAKA
action-status (local-exec): !!!! image_name: OPNsense 18.1 - 20180717Z102528
action-status (local-exec): !!!! image_action_outfile: /tmp/opnsense-YDYAKA-image-action.json
action-status (local-exec): !!!!
action-status (local-exec): !!!! Remember to terraform destroy resources once image action is complete
action-status (local-exec): !!!!

The user should perform a terraform destroy once complete to remove the resources that have allocated in the local tfstate - they can all safely be destroyed, the new AMI will not be removed in this destroy action because the action to create the image is performed as a local-exec call via awscli thus preventing it from being a Terraform resource.

Use your new AMI

You are now able to start a new AWS EC2 instance using the AMI that has been created for you. Your new AMI will be listed under the "My AMIs" section when you choose an Amazon Machine Image to start within the AWS console UI.

Warning!

The default rules used in this arrangement differ from the default OPNsense rules in that they allow access to the OPNsense control interfaces via TCP22 (SSH) and TCP443 (HTTPS) to facilitate your initial connection(s) to the system for setup etc.

Leaving the system in this arrangement is NOT recommended and you should take steps to restrict the source addresses that can connect to your OPNsense control interfaces.

Notes and Observations

  • The image "build" process leverages the OPNsense provided opnsense-bootstrap.sh tool to "convert" a FreeBSD Droplet into an OPNsense one, check it out here - https://github.com/opnsense/update
  • It is recommended that the user does not choose an instance size too small else the build will take a very long time, generally a t2.medium is appropriate for the build and will take approx 7 to 8 minutes.
  • You will see a lot of Terraform log output as the build process continues, keep in mind that builds can fail for many surprising reasons, external packages may not download and kernel-panics have been observed, so it is worth keeping an eye on the Terraform logging output to make sure nothing really obvious is going wrong.
  • Remember to issue the terraform destroy at the end, else you may become confused what state you are in the next time to come to roll another Droplet based OPNsense image.

What about Packer?

Packer, also produced by Hashicorp is an awesome tool, but requires learning yet another tool-chain. Since the resulting Digital Ocean images are targeted at DevOps people that use Terraform, it just felt more natural to do the whole build process in Terraform.

Builds Confirmed

  • (v0.3.0) amazon-ami: FreeBSD 11.2-STABLE-amd64-2018-07-09 (ami-36a0dedc) > OPNsense 18.1.12 (@ 2018-07-17T09:09:00Z)
  • (v0.3.1) amazon-ami: FreeBSD 11.2-STABLE-amd64-2018-08-02 (ami-285012c2) > OPNsense 18.7.0 (@ 2018-08-05T15:12:13Z)

Compute Providers Supported


Input Variables - Required

opnsense_release

The OPNsense release to target for this image build

digitalocean_region

The DigitalOcean region-slug to start this digitalocean-droplet within (nyc1, sgp1, lon1, nyc3, ams3, fra1, tor1, sfo2, blr1)

digitalocean_token

Your DigitalOcean API token used to issue cURL API calls directly to DigitalOcean to create the required image

Input Variables - Optional

root_passwd

The initial root password for OPNsense once the image is built.

  • default = "opnsense"

hostname

The hostname applied to this digitalocean-droplet within the image build process only.

  • default = "opnsense-image"

digitalocean_image

The DigitalOcean image to use as the base for this digitalocean-droplet.

  • default = "freebsd-11-1-x64"

digitalocean_size

The size to use for this digitalocean-droplet.

  • default = "s-1vcpu-1gb"

digitalocean_ipv6

Enable/disable getting a public IPv6 on this digitalocean-droplet.

  • default = true

digitalocean_private_networking

Enable/disable private-networking functionality on this digitalocean-droplet.

  • default = true

do_opnsense_install

Cause OPNsense to be installed once the instance is reachable.

  • default = 1

do_cleanup_shutdown

Cause the system to perform cleanup operations and then shutdown.

  • default = 1

do_image

Cause a Digital Ocean Droplet image to be taken of the Droplet while powered off.

  • default = 1

do_self_destruct

Cause the Droplet that was used to create the snapshot image to delete itself itself once the image is done.

  • default = 1

Outputs

provider

The compute provider name.

region

The compute provider region identifier.

build_id

The build identifier used to generate this image.

image_name

The image name given to this volume.

image_action_outfile

The output file from the image action call to the compute provider.


Authors

Module managed by Verb Networks.

License

Apache 2 Licensed. See LICENSE file for full details.