Replace Terraform with OpenTofu

.ansible-lint

@@ -5,6 +5,5 @@ skip_list:
 # Roles and modules imported from https://opendev.org/zuul/zuul-jobs
 mock_roles:
   - ensure-pip
-  - ensure-terraform
 mock_modules:
   - zuul_return

.github/workflows/check-terraform-syntax.yml → .github/workflows/check-opentofu-syntax.yml

@@ -1,18 +1,18 @@
 ---
-name: Check terraform syntax
+name: Check opentofu syntax
 
 "on":
   push:
     paths:
       - 'terraform/**'
-      - '.github/workflows/check-terraform-syntax.yml'
+      - '.github/workflows/check-opentofu-syntax.yml'
   pull_request:
     paths:
       - 'terraform/**'
-      - '.github/workflows/check-terraform.yml'
+      - '.github/workflows/check-opentofu-syntax.yml'
 
 jobs:
-  check-terraform-syntax:
+  check-opentofu-syntax:
     runs-on: ubuntu-latest
 
     steps:
@@ -24,18 +24,18 @@ jobs:
       - name: create credentials
         run: cp secure.yaml.sample secure.yaml
         working-directory: ./terraform
-      - name: Setup terraform
-        uses: hashicorp/setup-terraform@v2
+      - name: Setup OpenTofu
+        uses: opentofu/setup-opentofu@v1
         with:
-          terraform_version: 1.4.6  # renovate: datasource=github-releases depName=hashicorp/terraform
-      - name: Terraform init
-        run: terraform init
+          tofu_version: 1.6.0  # renovate: datasource=github-releases depName=opentofu/opentofu
+      - name: Tofu init
+        run: tofu init
         working-directory: ./terraform
-      - name: Terraform validate
-        run: terraform validate
+      - name: Tofu validate
+        run: tofu validate
         working-directory: ./terraform
         env:
           ENVIRONMENT: gx-betacloud
-      - name: Terraform format
-        run: terraform fmt -check
+      - name: Tofu format
+        run: tofu fmt -check
         working-directory: ./terraform

doc/Upgrade-Guide.md

@@ -151,7 +151,7 @@ on the cluster unless you have changed its configuration by tweaking
 `cluster-template.yaml` (which you almost never do!) or `clusterctl.yaml`
 (which you do often).
 
-The other template file that changed -- however, some terraform logic is used to
+The other template file that changed -- however, some opentofu logic is used to
 prefill it with values. So you can't copy it from git.
 
 ##### R2 to R3

doc/application-credentials.md

@@ -1,6 +1,6 @@
 # Application Credentials
 
-The terraform creates an [application credential](https://docs.openstack.org/keystone/wallaby/user/application_credentials.html) that it passes into the created VM. This one is then used to authenticate the cluster API provider against the OpenStack API to allow it to create resources needed for the k8s cluster.
+OpenTofu creates an [application credential](https://docs.openstack.org/keystone/wallaby/user/application_credentials.html) that it passes into the created VM. This one is then used to authenticate the cluster API provider against the OpenStack API to allow it to create resources needed for the k8s cluster.
 
 The AppCredential has a few advantages:
 

doc/continuous-integration.md

@@ -80,7 +80,7 @@ This section describes Zuul jobs defined within the k8s-cluster-api-provider pro
     availability in Zuul worker node. Parent job also defines a Zuul semaphore `semaphore-openstack-access`,
     that ensures that only one `openstack-access-base` job (or its children) can run at a time
   - See a high level `k8s-cluster-api-provider-e2e-conformance` job steps:
-    - Pre-run playbook `dependencies.yaml` installs project prerequisites, e.g. terraform, yq, etc. 
+    - Pre-run playbook `dependencies.yaml` installs project prerequisites, e.g. opentofu, yq, etc.
     - Main playbook `e2e.yaml` spawns a k8s cluster, runs sonobuoy conformance test, and cleans created infrastructure, all by k8s-cluster-api-provider scripts
     - Cleanup-run playbook `cleanup.yaml` runs `ospurge`, cleanup created application credentials and keypair to ensure that multiple e2e runs do not interfere
 
@@ -90,7 +90,7 @@ This section describes Zuul jobs defined within the k8s-cluster-api-provider pro
     availability in Zuul worker node. Parent job also defines a Zuul semaphore `semaphore-openstack-access`,
     that ensures that only one `openstack-access-base` job (or its children) can run at a time
   - See a high level `k8s-cluster-api-provider-e2e-quick` job steps:
-    - Pre-run playbook `dependencies.yaml` installs project prerequisites, e.g. terraform, yq, etc. 
+    - Pre-run playbook `dependencies.yaml` installs project prerequisites, e.g. opentofu, yq, etc.
     - Main playbook `e2e.yaml` spawns a k8s cluster, runs sonobuoy quick test, and cleans created infrastructure, all by k8s-cluster-api-provider scripts
     - Cleanup-run playbook `cleanup.yaml` runs `ospurge`, cleanup created application credentials and keypair to ensure that multiple e2e runs do not interfere
 

doc/make-reference.md

@@ -86,13 +86,13 @@ This will deploy the [cluster-stack-operator](https://github.com/SovereignCloudS
 ### Make clean
 
 ``make clean`` does ssh to the capi management server to clean up the created clusters prior
-to terraform cleaning up the resources it has created. This is sometimes insufficient to clean up
+to opentofu cleaning up the resources it has created. This is sometimes insufficient to clean up
 unfortunately, some error in the deployment may result in resources left around.
 
 ### Make fullclean
 
 ``make fullclean`` uses a custom script `cleanup/cleanup.sh` (using the openstack CLI) to clean up
-everything while trying to not hit any resources not created by the CAPI or terraform for
+everything while trying to not hit any resources not created by the CAPI or opentofu for
 clusters from this management host.
 It is the recommended way for doing cleanups if ``make clean`` fails. Watch out for leftover
 floating IP addresses and persistent volumes, as these can not be easily traced back to the
@@ -105,7 +105,7 @@ case we don't need the force options.
 
 You can purge the whole project via ``make purge``. Be careful with that command as it will purge
 *all resources in the OpenStack project* even those that have not been created through this
-Terraform script or the Cluster API.
+OpenTofu script or the Cluster API.
 It requires the [``ospurge``](https://opendev.org/x/ospurge) tool.
 Install it with ``python3 -m pip install git+https://git.openstack.org/openstack/ospurge``.
 
@@ -145,52 +145,52 @@ This will run tests of the configuration on testcluster using sonobuoy of the st
 
 This will run tests of the configuration on testcluster using sonobuoy of the CSI.
 
-## Terraform commands
+## OpenTofu commands
 
 ### make init
 
 ``make init``
 
-This will initialize terraform. It will download the required providers and modules.
+This will initialize opentofu. It will download the required providers and modules.
 It will also select or create a new workspace for you. The workspace name is the same as the
 ``ENVIROMENT`` variable.
 
 ### make attach
 
 ``make attach RESOURCE=<resource-id> PARAMS=...``
 
-This will attach a resource to the terraform state. This is useful if you have created a resource outside of terraform
-and want to manage it with terraform.
+This will attach a resource to the opentofu state. This is useful if you have created a resource outside of opentofu
+and want to manage it with opentofu.
 
 ### make detach
 
 ``make detach RESOURCE=<resource-id> PARAMS=...``
 
-This will detach a resource from the terraform state. This is useful if you have changed a resource outside of terraform
-or you no longer want to manage it with terraform.
+This will detach a resource from the opentofu state. This is useful if you have changed a resource outside of opentofu
+or you no longer want to manage it with opentofu.
 
 ### make state-push
 
 ``make state-push``
 
-This will push the terraform state to specified storage if set. This is useful if you don't want to store the state
+This will push the opentofu state to specified storage if set. This is useful if you don't want to store the state
 locally.
 
 ### make dry-run
 
 ``make dry-run``
 
-This will run a dry-run of the terraform apply command. This is useful if you want to see what terraform will do before
+This will run a dry-run of the opentofu apply command. This is useful if you want to see what opentofu will do before
 actually doing it.
 
 ### make show
 
 ``make show``
 
-This will show the terraform state. This is useful if you want to see what terraform is managing.
+This will show the opentofu state. This is useful if you want to see what opentofu is managing.
 
 ### make list
 
 ``make list``
 
-This will list all the resources managed by terraform. This is useful if you want to see what terraform is managing.
+This will list all the resources managed by opentofu. This is useful if you want to see what opentofu is managing.

doc/overview.md

@@ -13,6 +13,6 @@ Once we as the SCS Community have the gitops style cluster control working, the
 
 The provided solution covers the following two main topics:
 
-1. Automation (terraform, Makefile) to bootstrap a cluster-API management server by installing kind on a vanilla Ubuntu image and deploying some tools on this node ([kubectl](https://kubernetes.io/docs/reference/kubectl/overview/), [openstack CLI tools](https://docs.openstack.org/newton/user-guide/common/cli-install-openstack-command-line-clients.html), [k9s](https://github.com/derailed/k9s), [cilium](https://cilium.io/), [calico](https://www.tigera.io/tigera-products/calico/), [helm](https://helm.sh/), [flux](https://fluxcd.io/) ...) and deploying [cluster-API](https://cluster-api.sigs.k8s.io/) (clusterctl) and the [OpenStack cluster-api provider](https://github.com/kubernetes-sigs/cluster-api-provider-openstack) along with suitable credentials. The terraform automation is driven by a Makefile for convenience. The tooling also contains all the logic to clean up again. The newly deployed node clones this git repository early in the bootstrap process and uses the thus received files to set up the management cluster and scripts.
+1. Automation (opentofu, Makefile) to bootstrap a cluster-API management server by installing kind on a vanilla Ubuntu image and deploying some tools on this node ([kubectl](https://kubernetes.io/docs/reference/kubectl/overview/), [openstack CLI tools](https://docs.openstack.org/newton/user-guide/common/cli-install-openstack-command-line-clients.html), [k9s](https://github.com/derailed/k9s), [cilium](https://cilium.io/), [calico](https://www.tigera.io/tigera-products/calico/), [helm](https://helm.sh/), [flux](https://fluxcd.io/) ...) and deploying [cluster-API](https://cluster-api.sigs.k8s.io/) (clusterctl) and the [OpenStack cluster-api provider](https://github.com/kubernetes-sigs/cluster-api-provider-openstack) along with suitable credentials. The opentofu automation is driven by a Makefile for convenience. The tooling also contains all the logic to clean up again. The newly deployed node clones this git repository early in the bootstrap process and uses the thus received files to set up the management cluster and scripts.
 
 2. This node can be connected to via ssh and the deployed scripts there can be used to manage workload clusters and then deploy various standardized tools (such as e.g. [OpenStack Cloud Controller Manager](https://github.com/kubernetes/cloud-provider-openstack)(OCCM), [cinder CSI](https://github.com/kubernetes/cloud-provider-openstack/blob/master/docs/cinder-csi-plugin/using-cinder-csi-plugin.md), calico or cilium CNI, [nginx ingress controller](https://kubernetes.github.io/ingress-nginx/), [cert-manager](https://cert-manager.io/), ...) and run tests (e.g. CNCF conformance with [sonobuoy](https://sonobuoy.io/)). The tools and artifacts can be updated via `git pull` at any time and the updated settings rolled out to the workload clusters. Note that the script collection will eventually be superseded by the [capi-helm-charts](https://github.com/stackhpc/capi-helm-charts). The medium-term goal is to actually create a reconciliation loop here that would perform life-cycle-management for clusters according to the cluster configuration stored in an enhanced [cluster-api style](https://cluster-api.sigs.k8s.io/clusterctl/configuration.html) clusterctl.yaml from git repositories and thus allow a pure [gitops](https://www.weave.works/technologies/gitops/) style cluster management without ever ssh'ing to the management server.

doc/quickstart.md

@@ -7,7 +7,7 @@ via [cluster-api](https://cluster-api.sigs.k8s.io/)(CAPI).
 
 - make
 - kubectl
-- terraform
+- opentofu
 - yq
 - python3-openstackclient, python3-octaviaclient
 

doc/requirements.md

@@ -12,12 +12,12 @@ Linux/MacOS/WSL laptop.
 Requirements for the deployment host:
 
 - You need to have installed:
-    - Terraform (<https://learn.hashicorp.com/tutorials/terraform/install-cli>).
+    - OpenTofu (<https://opentofu.org/docs/intro/install/>).
     - `yq` (python3-yq or yq snap)
     - GNU make
     - openstack (python3-openstackclient) and plugin for octavia (python3-octaviaclient) Via pip or your distribution.
       *Needed only in case you want to clean the management server or interact with openstack directly.*
-- You must have credentials to access the cloud. Terraform will look for `clouds.yaml` and optionally `secure.yaml` in
+- You must have credentials to access the cloud. OpenTofu will look for `clouds.yaml` and optionally `secure.yaml` in
   the current working directory (`terraform`), in `~/.config/openstack/` or `/etc/openstack` (in this order), just like
   the [openstack client](https://docs.openstack.org/python-openstackclient/latest/configuration/index.html#clouds-yaml).
 - The API endpoints of the OpenStack cloud should have a certificate signed by a trusted CA. (Self-signed or custom CAs

doc/usage/containter-registry-configuration.md

@@ -18,7 +18,7 @@ This feature could be useful in various uses cases, e.g.:
 SCS KaaS reference implementation allows users to pass container registry host config
 files to configure containerd. Containerd configuration is applied cluster wide as it
 is CRI used in SCS KaaS clusters. Additional registry host config files for containerd
-could be passed through the `containerd_registry_files` terraform variable. This variable
+could be passed through the `containerd_registry_files` opentofu variable. This variable
 expects an object with two attributes:
 
 - `hosts` attribute defines additional registry host config files for containerd.
@@ -33,11 +33,11 @@ The default value of the `containerd_registry_files` variable instructs containe
 container registry, see related
 issue [#417](https://github.com/SovereignCloudStack/k8s-cluster-api-provider/issues/417).
 The host config file used as a default is
-defined [here](https://github.com/SovereignCloudStack/k8s-cluster-api-provider/blob/1b6ef9d4c64c94bc77144a072e0309d484de54be/terraform/files/containerd/docker.io).  
+defined [here](https://github.com/SovereignCloudStack/k8s-cluster-api-provider/blob/main/terraform/files/containerd/docker.io).
 This should prevent issues with pull rate limiting from DockerHub public container registry,
 e.g. [#414](https://github.com/SovereignCloudStack/k8s-cluster-api-provider/issues/414).
 
-The above default value could be overridden using any techniques that Terraform allows, e.g.
+The above default value could be overridden using any techniques that OpenTofu allows, e.g.
 via environment variable:
 
 ```bash
@@ -80,7 +80,7 @@ upgrade existing environments (workload clusters) with stable releases which are
 tested.
 
 This tutorial instructs (advanced) users what needs to be done on the cluster management
-instance without re-deploying it via Terraform. Patches described in this tutorial
+instance without re-deploying it via OpenTofu. Patches described in this tutorial
 modify existing environment with changes that have been developed within "containerd registry"-related PRs:
 - PR#432
 - PR#447

doc/usage/custom-ca.md

@@ -28,7 +28,7 @@ Here the file `ca-bundle.pem` contains custom root CA and potentially intermedia
 Steps of what happens with the custom cacert in k8s-cluster-api-provider:
 1. `cacert` setting is provided inside clouds.yaml
 2. Cacert file referenced by `cacert` key (1.) is copied to the management server
-   directory `~/cluster-defaults/${cloud_provider}-cacert` by Terraform
+   directory `~/cluster-defaults/${cloud_provider}-cacert` by OpenTofu
 3. During the management server bootstrap process cacert is injected to
    the *~/cluster-defaults/cluster-template.yaml* to *KubeadmControlPlane* and *KubeadmConfigTemplate* files
    as file with cacert content from already defined secret *${CLUSTER_NAME}-cloud-config* and will be later

doc/usage/harbor.md

@@ -4,7 +4,7 @@ Project [k8s-harbor](https://github.com/SovereignCloudStack/k8s-harbor) is used
 container registry deployment https://registry.scs.community/.
 
 With this project, it is possible to deploy Harbor in a similar way into the workload cluster.
-If you want to deploy Harbor, set terraform variable `deploy_harbor = true`. It will deploy
+If you want to deploy Harbor, set opentofu variable `deploy_harbor = true`. It will deploy
 Harbor with [default options](#default-deployment). The recommended approach is to
 set up also [persistence](#persistence) and [ingress with tls](#ingress-and-tls).
 

playbooks/dependencies.yaml

@@ -7,17 +7,18 @@
     kubectl_version: "1.28.1"
     openstackclient_version: "6.2.0"
     octaviaclient_version: "3.4.0"
-    tf_version: "1.4.6"
+    opentofu_version: "1.6.0"
     install_dir: "{{ ansible_user_dir }}/.local/bin"
   environment:
     PATH: "{{ install_dir }}:{{ ansible_env.PATH }}"
   roles:
   - role: ensure-pip  # https://zuul-ci.org/docs/zuul-jobs/latest/python-roles.html#role-ensure-pip
-  - role: ensure-terraform  # https://zuul-ci.org/docs/zuul-jobs/latest/hashicorp-roles.html#role-ensure-terraform
-    vars:
-      terraform_version: "{{ tf_version }}"
-      terraform_install_dir: "{{ install_dir }}"
   tasks:
+  - name: Make sure installation directory exists
+    ansible.builtin.file:
+      path: "{{ install_dir }}"
+      state: directory
+      mode: 0755
   - name: Install jq
     ansible.builtin.get_url:
       url: "https://github.com/jqlang/jq/releases/download/jq-{{ jq_version }}/jq-linux64"
@@ -44,6 +45,46 @@
       extra_args: --user
   - name: Install ip command
     ansible.builtin.package:
-      name:
-        - iproute2
+      name: iproute2
     become: true
+  - name: Install OpenTofu
+    # TODO: write separate role ensure-opentofu, see https://opendev.org/zuul/zuul-jobs/src/branch/master/roles/ensure-terraform
+    block:
+      - name: Install unzip
+        ansible.builtin.package:
+          name: unzip
+        become: true
+      - name: Create temp directory
+        ansible.builtin.tempfile:
+          state: directory
+        register: tofu_install_tempdir
+      - name: Download OpenTofu archive
+        ansible.builtin.get_url:
+          url: "https://github.com/opentofu/opentofu/releases/download/v{{ opentofu_version }}/tofu_{{ opentofu_version }}_linux_amd64.zip"
+          dest: "{{ tofu_install_tempdir.path }}/opentofu.zip"
+          mode: 0644
+      # TODO: checksum
+      - name: Create OpenTofu package directory
+        ansible.builtin.file:
+          path: "{{ tofu_install_tempdir.path }}/opentofu"
+          state: directory
+          mode: 0755
+      - name: Unarchive OpenTofu
+        ansible.builtin.unarchive:
+          src: "{{ tofu_install_tempdir.path }}/opentofu.zip"
+          dest: "{{ tofu_install_tempdir.path }}/opentofu"
+          remote_src: true
+      - name: Install OpenTofu
+        ansible.builtin.copy:
+          src: "{{ tofu_install_tempdir.path }}/opentofu/tofu"
+          dest: "{{ install_dir }}/tofu"
+          mode: "+x"
+          remote_src: true
+      - name: Remove temp directory
+        ansible.builtin.file:
+          path: "{{ tofu_install_tempdir }}"
+          state: absent
+      - name: Output OpenTofu version
+        ansible.builtin.command: "tofu version"
+        register: tofu_version
+        changed_when: tofu_version.rc != 0

playbooks/e2e.yaml

@@ -82,7 +82,7 @@
         executable: /bin/bash
       register: mtu
       changed_when: false
-    - name: Set mtu for the default interface  # workaround for terraform file provisioner scp error
+    - name: Set mtu for the default interface  # workaround for opentofu file provisioner scp error
       ansible.builtin.command: "ip link set {{ dev.stdout }} mtu 1500"
       become: true
       when: mtu.stdout != "1500"