# Terraform and OpenTofu Provider

The ZCP provider lets Terraform and OpenTofu manage instances, networks, Kubernetes, DNS, storage, and related platform resources. The [v1.0.0](https://github.com/zsoftly/terraform-provider-zcp/releases/tag/v1.0.0) provider uses the CLI v1.0.1 resource model.

## Install the Provider

Use the registry address that matches your IaC tool. Replace the version constraint when you choose a later release.

```hcl
# OpenTofu
terraform {
  required_providers {
    zcp = {
      source  = "registry.opentofu.org/zsoftly/zcp"
      version = "~> 1.0.0"
    }
  }
}
```

```hcl
# Terraform
terraform {
  required_providers {
    zcp = {
      source  = "registry.terraform.io/zsoftly/zcp"
      version = "~> 1.0.0"
    }
  }
}
```

Set the token outside version control. The provider reads `ZCP_BEARER_TOKEN`, `ZCP_API_URL`, and `ZCP_PROJECT` from the environment. `ZCP_API_URL` is optional and defaults to `https://api.zcp.zsoftly.ca/api`. Set it only when using a different API endpoint.

```bash
export ZCP_BEARER_TOKEN="your-token"
export ZCP_PROJECT="your-project-slug"
```

## Upgrade Existing Configuration

Update your provider version constraint, then run the matching commands from the directory that contains your configuration:

```bash
terraform init -upgrade
terraform plan
```

```bash
tofu init -upgrade
tofu plan
```

Review the plan before applying it. The migration notes below identify replacements that version 1.0.0 can propose.

## Named Compute and Root Storage

New instances require a named compute plan, a named root-storage plan, and root-disk capacity. The provider rejects custom CPU, memory, and disk inputs for new instances. Query plan values available to your account with the CLI before writing configuration.

Supply the referenced `var.*` values through your variables or module inputs.

```hcl
resource "zcp_instance" "web" {
  name              = "web"
  cloud_provider    = var.cloud_provider_slug
  region            = var.region_slug
  template          = var.template_slug
  plan              = var.compute_plan_slug
  blockstorage_plan = var.root_storage_plan_slug
  root_disk_size    = 100
  billing_cycle     = "hourly"
  network           = var.network_slug
  storage_category  = var.storage_category_slug
}
```

`blockstorage_plan` and `root_disk_size` force replacement when changed. Existing custom-plan state remains readable and destroyable, but a new or replacement instance must use named plans.

## Kubernetes Cluster Migration

New clusters use separate fixed plans for control-plane nodes, worker nodes, and root storage. The root-disk size applies to every cluster node.

Supply the `var.*` values for your cluster configuration.

```hcl
resource "zcp_kubernetes_cluster" "main" {
  name               = "main"
  cloud_provider     = var.cloud_provider_slug
  region             = var.region_slug
  version            = var.kubernetes_version
  control_plane_plan = var.control_plane_plan_slug
  worker_plan        = var.worker_plan_slug
  storage_plan       = var.root_storage_plan_slug
  root_disk_size     = 100
  billing_cycle      = "hourly"
  workers            = 3
  storage_category   = var.storage_category_slug
  ssh_key            = var.ssh_key_name
}
```

The legacy `plan` field remains only for existing state. Switching an existing cluster to the separate plan shape replaces the cluster because the API has no role-specific plan update operation. Review `terraform plan` or `tofu plan` before applying. Cluster cancellation can leave a network you must delete separately.

The provider records requested Kubernetes autoscaling bounds. It does not confirm an immediate worker increase, and automatic scale-out remains unverified.

## VM Backup Schedules

`zcp_vm_backup_schedule` manages a recurring VM backup policy, including its interval, time zone, retention, and paused state.

```hcl
resource "zcp_vm_backup_schedule" "web" {
  virtual_machine = zcp_instance.web.id
  interval        = "dailyAt"
  at              = "02:00"
  timezone        = "America/Toronto"
  retention       = 7
  paused          = false
  region          = var.region_slug
}
```

Use `hourly`, `dailyAt`, `weeklyOn`, or `monthlyOn`. The provider rejects `everyOtherDay` before writing a policy because the platform can persist that interval and then report an error.

## Other Version 1.0.0 Changes

- `zcp_volume` accepts a named storage `plan` with `size` for capacity in GB.
- DNS domain lookups retrieve pages until the requested slug is found.
- Updating an instance `name` changes its hostname and can restart a running instance or start a stopped instance.
- The provider does not store kubeconfig or cluster credentials.

See the [provider source and generated reference](https://github.com/zsoftly/terraform-provider-zcp) for the complete schema and import formats. See also: [CLI Reference](https://docs.zcp.zsoftly.ca/public-cloud/cli/reference), [API Authentication](https://docs.zcp.zsoftly.ca/public-cloud/api/authentication).