ci: Add test plan that runs CI tests and customize it for each role (#133)

* Calculate number of managed nodes with this formula:
    (( number_of_test_playbooks / 10 + 1 ))
* Add README explaining how to run the plan locally and remotely

Signed-off-by: Sergei Petrosian <spetrosi@redhat.com>

Author: spetrosi

.codespellrc

@@ -5,4 +5,6 @@ check-hidden = true
 ignore-multiline-regex = codespell:ignore-begin.*codespell:ignore-end
 ignore-words = .codespell_ignores
 # skip-file is not available https://github.com/codespell-project/codespell/pull/2759
+# .pandoc_template.html5 contains a typo in Licence that we shouldn't edit
+# .README.html is generated from README.md automatically - no need to check spelling
 skip = .pandoc_template.html5,.README.html

.github/workflows/tft.yml

@@ -157,7 +157,6 @@ jobs:
         uses: sclorg/testing-farm-as-github-action@v3
         if: contains(needs.prepare_vars.outputs.supported_platforms, matrix.platform)
         with:
-          git_url: https://github.com/linux-system-roles/tft-tests
           git_ref: main
           pipeline_settings: '{ "type": "tmt-multihost" }'
           environment_settings: '{ "provisioning": { "tags": { "BusinessUnit": "system_roles" } } }'
@@ -180,8 +179,7 @@ jobs:
           tf_scope: private
           api_key: ${{ secrets.TF_API_KEY_RH }}
           update_pull_request_status: false
-          tmt_hardware: '{ "memory": ">= ${{ needs.prepare_vars.outputs.memory }} MB" }'
-          tmt_plan_filter: "tag:general,bootloader"
+          tmt_plan_filter: "tag:playbooks_parallel,bootloader"
 
       - name: Set final commit status
         uses: myrotvorets/set-commit-status-action@master

plans/README-plans.md

@@ -0,0 +1,55 @@
+# Introduction CI Testing Plans
+
+Linux System Roles CI runs [tmt](https://tmt.readthedocs.io/en/stable/index.html) test plans in [Testing farm](https://docs.testing-farm.io/Testing%20Farm/0.1/index.html) with the [tft.yml](https://github.com/linux-system-roles/bootloader/blob/main/.github/workflows/tft.yml) GitHub workflow.
+
+The `plans/test_playbooks_parallel.fmf` plan is a test plan that runs test playbooks in parallel on multiple managed nodes.
+`plans/test_playbooks_parallel.fmf` is generated centrally from `https://github.com/linux-system-roles/.github/`.
+The automation calculates the number of managed nodes to provision with this formula:
+
+```plain
+number-of-test-playbooks / 10 + 1
+```
+
+The `plans/test_playbooks_parallel.fmf` plan does the following steps:
+
+1. Provisions systems to be used as a control node and as managed nodes.
+2. Does the required preparation on systems.
+3. For the given role and the given PR, runs the general test from [test.sh](https://github.com/linux-system-roles/tft-tests/blob/main/tests/general/test.sh).
+
+The [tft.yml](https://github.com/linux-system-roles/bootloader/blob/main/.github/workflows/tft.yml) workflow runs the above plan and uploads the results to our Fedora storage for public access.
+This workflow uses Testing Farm's Github Action [Schedule tests on Testing Farm](https://github.com/marketplace/actions/schedule-tests-on-testing-farm).
+
+## Running Tests
+
+You can run tests locally with the `tmt try` cli or remotely in Testing Farm.
+
+### Running Tests Locally
+
+1. Install `tmt` as described in [Installation](https://tmt.readthedocs.io/en/stable/stories/install.html).
+2. Change to the role repository directory.
+3. Modify `plans/test_playbooks_parallel.fmf` to suit your requirements:
+    1. Due to [issue #3138](https://github.com/teemtee/tmt/issues/3138), comment out all managed nodes except for one.
+    2. Optionally modify environment variables to, e.g. run only specified test playbooks by modifying `SYSTEM_ROLES_ONLY_TESTS`.
+4. Enter `tmt try -p plans/test_playbooks_parallel <platform>`.
+    This command identifies the `plans/test_playbooks_parallel.fmf` plan and provisions local VMs, a control node and a managed node.
+5. `tmt try` is in development and does not identify tests from URL automatically, so after provisioning the machines, you must type `t`, `p`, `t` from the interactive prompt to identify tests, run preparation steps, and run the tests.
+
+### Running in Testing Farm
+
+1. Install `testing-farm` as described in [Installation](https://gitlab.com/testing-farm/cli/-/blob/main/README.adoc#user-content-installation).
+2. Change to the role repository directory.
+3. If you want to run tests with edits in your branch, you need to commit and push changes first to some branch.
+4. You can uncomment "Inject your ssh public key to test systems" discover step in the plan if you want to troubleshoot tests by SSHing into test systems in Testing Farm.
+5. Enter `testing-farm request`.
+    Edit to your needs.
+
+    ```bash
+    $ TESTING_FARM_API_TOKEN=<your_api_token> \
+        testing-farm request --pipeline-type="tmt-multihost" \
+        --plan-filter="tag:playbooks_parallel" \
+        --git-url "https://github.com/<my_user>/bootloader" \
+        --git-ref "<my_branch>" \
+        --compose CentOS-Stream-9 \
+        -e "SYSTEM_ROLES_ONLY_TESTS=tests_default.yml" \
+        --no-wait
+    ```

plans/test_playbooks_parallel.fmf

@@ -0,0 +1,69 @@
+summary: A general test for a system role
+tag: playbooks_parallel
+provision:
+  # TF uses `how: artemis`, and `tmt try`` uses `how: virtual`.
+  # Hence there is no need to define `how` explicitly.
+  - name: control-node1
+    role: control_node
+    # `connection: system` is required for `how: virtual` to assign VMs a real
+    # IP making SSH configuration easier.
+    # This setting is ignored in `artemis`, so we can leave it as is.
+    connection: system
+  - name: managed-node1
+    role: managed_node
+    connection: system
+environment:
+  ANSIBLE_VER: 2.17
+  REPO_NAME: bootloader
+  PYTHON_VERSION: 3.12
+  SYSTEM_ROLES_ONLY_TESTS: ""  # tests_default.yml
+  TEST_LOCAL_CHANGES: true
+  PR_NUM: ""
+  LINUXSYSTEMROLES_USER: ""
+  LINUXSYSTEMROLES_DOMAIN: ""
+  LINUXSYSTEMROLES_SSH_KEY: ""
+  ARTIFACTS_DIR: ""
+  ARTIFACTS_URL: ""
+  LSR_TFT_DEBUG: false
+prepare:
+  - name: Use vault.centos.org repos (CS 7, 8 EOL workaround)
+    script: |
+      if grep -q 'CentOS Stream release 8' /etc/redhat-release; then
+        sed -i '/^mirror/d;s/#\(baseurl=http:\/\/\)mirror/\1vault/' /etc/yum.repos.d/*.repo
+      fi
+      if grep -q 'CentOS Linux release 7.9' /etc/redhat-release; then
+        sed -i '/^mirror/d;s/#\?\(baseurl=http:\/\/\)mirror/\1vault/' /etc/yum.repos.d/*.repo
+      fi
+  # Replace with feature: epel: enabled once https://github.com/teemtee/tmt/pull/3128 is merged
+  - name: Enable epel to install beakerlib
+    script: |
+      # CS 10 and Fedora doesn't require epel
+      if grep -q -e 'CentOS Stream release 10' -e 'Fedora release' /etc/redhat-release; then
+        exit 0
+      fi
+      yum install epel-release yum-utils -y
+      yum-config-manager --enable epel epel-debuginfo epel-source
+discover:
+  - name: Prepare managed node
+    how: fmf
+    where: managed_node
+    filter: tag:prep_managed_node
+    url: https://github.com/linux-system-roles/tft-tests
+    ref: main
+  - name: Run test playbooks from control_node
+    how: fmf
+    where: control_node
+    filter: tag:test_playbooks
+    url: https://github.com/linux-system-roles/tft-tests
+    ref: main
+  # Uncomment this step for troubleshooting
+  # This is required because currently testing-farm cli doesn't support running multi-node plans
+  # You can set ID_RSA_PUB in the environment section above to your public key to distribute it to nodes
+  # - name: Inject your ssh public key to test systems
+  #   how: fmf
+  #   where: control_node
+  #   filter: tag:reserve_system
+  #   url: https://github.com/linux-system-roles/tft-tests
+  #   ref: main
+execute:
+    how: tmt