terraform-provider-k8s is a Terraform provider to manage Kubernetes resources.

• Uses the Kubernetes client-go library to dynamically discover available resources
• Supports all Kubernetes resources, including alpha and beta versions
• Supports the latest version of Kubernetes
• Auto extract from live Kubernetes cluster and import as Terraform state
• Convert Kubernetes manifests to Terraform files
• A catalog of reusable modules
• A catalog of complete solutions using the modules
• Supports resources based on Custom Resource Definition(CRD).

• Docker
• kubeconfig file in the current directory
• Terraform 0.12
• Kubernetes v1.14 (Recommended for best CRD support)

docker pull registry.rebelsoft.com/terraform
alias tf='docker run -v `pwd`/kubeconfig:/kubeconfig -v `pwd`:/docker -w /docker --rm -it registry.rebelsoft.com/terraform terraform'
alias tfextract='docker run -v `pwd`/kubeconfig:/kubeconfig -v `pwd`:/docker -w /docker --rm -it registry.rebelsoft.com/terraform extractor'
alias tfgenerate='docker run -v `pwd`/kubeconfig:/kubeconfig -v `pwd`:/docker -w /docker --rm -it registry.rebelsoft.com/terraform generator'

For new projects or after install/upgrade, run tf init

After extracting existing resources to tf files, you can import the state for them.

Run tf import k8s_core_v1_service.nginx default.service.nginx to import, in this case a nginx service.

The resource format is resource "k8s_<group>_<version>_<kind>" "<name>".

Note the name here is the Terraform name and not the Kubernetes name.

All Kubernetes resources requires a name in the metadata like this

metadata {
  name = "nginx"
}

The set of supported groups, versions and kinds are loaded dynamically from your Kubernetes cluster.

The data source format is data "k8s_<group>_<version>_<kind>" "<name>".

The set of supported groups, versions and kinds are loaded dynamically from your Kubernetes cluster.

For example, this gets the cluster_ip of the httpbin service.

data "k8s_core_v1_service" "httpbin" {
  metadata {
    name = "httpbin"
  }
}
output "httpbin_cluster_ip" {
  value = "${data.k8s_core_v1_service.httpbin.spec.0.cluster_ip}"
}

The format of the resource Ids is <namespace>.<kind>.<name>. These are all Kubernetes identifiers.

If the Kind of the resource is not namespaced, such as Namespace resource, then the Id format is .<kind>.<name>.

Ids are used internally by Terraform to uniquely identify resource state. The user will only need to know them when importing state from a Kubernetes cluster.

Run tfgenerate to list all available resource and data source types

Run tfgenerate <resource> to generate a skeleton tf file for the resource. For example,

• tfgenerate k8s_core_v1_service to generate a Kubernetes Service
• tfgenerate k8s_apps_v1_deployment to generate a Kubernetes Deployment
• tfgenerate k8s_apps_v1_stateful_set to generate a Kubernetes StatefulSet

The tfextract will load the existing resources from YAML files and create Terraform files for them. One file per resource in the YAML file.

For local file, run the tfextract -filename <> -dir <target> command.

For remote file, run the tfextract -url <> -dir <target> command.

The tfextract will load the existing resources from Kubernetes and create Terraform files for them. One file per resource.

Run the tfextract -namespace <> -kind <> -name <> -import -dir <target> command. Any may be left blank but at least one must be set.

Example: tfextract -kind service -name nginx -dir <target> will extract the Service named nginx into a file called service-nginx.tf.

For this example, if -kind service was left blank then all resources named nginx will be extracted to their coresponding files. Likewise for -name nginx; if left blank then all services will be extracted.

The -import flag will automatically import the extracted resources as Terraform state.