custom-terraform-container

Container image with terraform, ibm terraform provider plugin, and other IBM Cloud dev tools installed

This document briefly outlines few ways to use the docker image created for the purpose of running terraform plans without having to install ibm terraform provider plugin and some other dev tools locally. You can spin-up a container using this image and directly begin using IBM Terraform provider plugin and provision IBM Cloud resources

Image on docker hub - custom-ibm-terraform

Provisioning a VM instance on IBM cloud using Terraform

terraform-provider-ibm

Use the image

Run using Docker

$ docker pull csphoenix/custom-ibm-terraform:0.11.14-0.21

File tf-variables.env looks something like the below. These variables are supplied to the sample terraform plan main.tf (part of the the image) that provisions a VSI with GPU. For details on how these env variables are used by terraform, take a look at Setting tf vars using environment variables

TF_VAR_hostname=tf-gpu-1
TF_VAR_domain=example.com
TF_VAR_public_vlan_id=12345
TF_VAR_private_vlan_id=56789
TF_VAR_os_reference_code=UBUNTU_18_64
TF_VAR_ibmcloud_api_key=<platform_apikey>
TF_VAR_iaas_classic_username=<infra_username>
TF_VAR_iaas_classic_api_key=<infra_apikey>

• Creating an API key
• Creating classic infrastructure API keys
• To obtain the vlan ids, you can use ibmcloud cli or ibmcloud console
  • $ ibmcloud login -apikey=<apikey>
  • $ ibmcloud sl vlan list.
  • If you don't have the ibm tools installed locally, you can use the console to obtain vlan ids. Navigate to vlans on your account and click on the vlan in the desired region, and use the ID number from the browser URL

$ docker run -it --rm --env-file=tf-variables.env csphoenix/custom-ibm-terraform:0.11.14-0.21

The above docker run command opens an interactive bash session. The container has terraform and ibm terraform providers installed. A sample terraform plan is also included in the image.

NOTE: If you run the script run-tf.sh as shown below, the terraform plan provided along with the image is executed and you might be billed accordingly. Take a look at the files before provisioning.

root@1ad9f9eddbcf:/workdir# ls
run-tf.sh  terraform-files
root@1ad9f9eddbcf:/workdir# ls terraform-files/
main.tf  variables.tf
root@1ad9f9eddbcf:/workdir# ./run-tf.sh 

The script run-tf.sh is wrapper script and is created for automating execution of various terraform commands like

terraform init
terraform plan
terraform apply

Instead of using the script, you can cd into the terraform-files directory and run the terraform commands as needed.

If the terraform apply is successful, it should show something like Apply complete! Resources: 1 added, 0 changed, 0 destroyed. Make sure to destroy it if it is created for testing purposes

root@1ad9f9eddbcf:/workdir# cd terraform-files/
root@1ad9f9eddbcf:/workdir/terraform-files# terraform destroy

Run on a kubernetes/openshift cluster

$ oc new-project terraform
$ oc adm policy add-scc-to-user anyuid -z default

$ kubectl run custom-ibm-terraform-container -it --rm --restart=Never --image=csphoenix/custom-ibm-terraform:0.11.14-0.21 
If you don't see a command prompt, try pressing enter.
root@custom-ibm-terraform-container:/workdir# 
root@custom-ibm-terraform-container:/workdir# ./run-tf.sh 

Because variable values are not supplied through env vars, terraform will prompt to enter values for variables that do not have default values set. Enter the values as needed and then the terraform apply will be executed. Make sure to do a destroy if created for testing

root@custom-ibm-terraform-container:/workdir# cd terraform-files
root@custom-ibm-terraform-container:/workdir/terraform-files# terraform destroy

Exit the pod

root@custom-ibm-terraform-container:/workdir# exit
exit
pod "custom-ibm-terraform-container" deleted

Points to be noted

• The main.tf in the container image is only a sample terraform plan that provisions a VSI with GPU. It can be modified to have more options or new terraform files can be copied into the container for testing
• The variables like the operating system, VSI GPU flavor, datacenter are set to default values and can be overidden. For details, take a look into variable.tf file
• The above steps show only one way to supply variable values to terrform plan. Other ways like using tfvars file or if running on a kube pod, a config map can be created and thus the variables can be retrieved as env variables. More details can be found at Input Variables

Build the image and run locally

├── Dockerfile
├── README.md
├── run-tf.sh
├── terraform-files
│   ├── main.tf
│   └── variables.tf
└── tf-variables.env

If you want to customize the image and have your own terraform plans in it, place your terraform files inside the terraform-files directory and re-build the image. The image can then be pushed to desired private registries.

Make sure you are in the root directory of the repo

$ docker build -t custom-ibm-terraform .

$ docker run -it --rm --env-file=tf-variables.env custom-ibm-terraform

Create a deployment on the cluster

• Sample deployment.yaml, secret.yaml are provided in the k8s_extra directory
• They can be used to create a deployment instead of creating temporary containers like above
• Note that the deployment is designed to pull provider credentials from a secret
• So, a secret has to be created first or you can change your deployment accordingly and use other methods to supply credentials

├── k8s_extra
│   ├── deployment.yaml
│   └── secret.yaml

$ kubectl create -f secret.yaml

$ kubectl create -f deployment.yaml

Then, you can exec into the create pod, copy new terraform plans and execute them as you like