Execute Vagrant cluster in CI (#57)

.github/workflows/test.yml

@@ -0,0 +1,69 @@
+---
+name: Test
+"on":
+  pull_request:
+  push:
+    branches:
+      - master
+
+jobs:
+  vagrant:
+    name: Vagrant
+    runs-on: macos-12
+
+    env:
+      HOMEBREW_NO_INSTALL_CLEANUP: 1
+      VAGRANT_CWD: ${{ github.workspace }}/vagrant
+
+    steps:
+      - name: Check out the codebase
+        uses: actions/checkout@v2
+
+      - name: Install Ansible
+        run: brew install ansible
+
+      - name: Install role dependencies
+        run: ansible-galaxy install -r collections/requirements.yml
+
+      - name: Configure VirtualBox
+        run: >-
+          sudo mkdir -p /etc/vbox &&
+          echo "* 192.168.30.0/24" | sudo tee -a /etc/vbox/networks.conf > /dev/null
+
+      - name: Cache Vagrant boxes
+        uses: actions/cache@v3
+        with:
+          path: |
+            ~/.vagrant.d/boxes
+          key: vagrant-boxes-${{ hashFiles('**/Vagrantfile') }}
+          restore-keys: |
+            vagrant-boxes
+
+      - name: Create virtual machines
+        run: vagrant up
+        timeout-minutes: 10
+
+      - name: Provision cluster using Ansible
+        # Since Ansible sets up _all_ machines, it is sufficient to run it only
+        # once (i.e, for a single node - we are choosing control1 here)
+        run: vagrant provision control1 --provision-with ansible
+        timeout-minutes: 25
+
+      - name: Set up kubectl on the host
+        run: brew install kubectl &&
+          mkdir -p ~/.kube &&
+          vagrant ssh control1 --command "cat ~/.kube/config" > ~/.kube/config
+
+      - name: Show cluster nodes
+        run: kubectl describe -A nodes
+
+      - name: Show cluster pods
+        run: kubectl describe -A pods
+
+      - name: Test cluster
+        run: $VAGRANT_CWD/test_cluster.py --verbose --locals
+        timeout-minutes: 5
+
+      - name: Destroy virtual machines
+        if: always() # do this even if a step before has failed
+        run: vagrant destroy --force

vagrant/Vagrantfile

@@ -3,12 +3,12 @@
 
 Vagrant.configure("2") do |config|
     # General configuration
-    config.vm.box = "generic/ubuntu2110"
+    config.vm.box = "generic/ubuntu2204"
     config.vm.synced_folder ".", "/vagrant", disabled: true
     config.ssh.insert_key = false
 
     config.vm.provider :virtualbox do |v|
-        v.memory = 4096
+        v.memory = 2048
         v.cpus = 2
         v.linked_clone = true
     end
@@ -50,7 +50,7 @@ Vagrant.configure("2") do |config|
           "master" => ["control1", "control2", "control3"],
           "node" => ["node1", "node2"],
           "k3s_cluster:children" => ["master", "node"],
-          "k3s_cluster:vars" => {"k3s_version" => "v1.23.4+k3s1",
+          "k3s_cluster:vars" => {"k3s_version" => "v1.24.3+k3s1",
                                  "ansible_user" => "vagrant",
                                  "systemd_dir" => "/etc/systemd/system",
                                  "flannel_iface" => "eth1",
@@ -58,9 +58,9 @@ Vagrant.configure("2") do |config|
                                  "k3s_token" => "supersecret",
                                  "extra_server_args" => "--node-ip={{ ansible_eth1.ipv4.address }} --flannel-iface={{ flannel_iface }} --no-deploy servicelb --no-deploy traefik",
                                  "extra_agent_args" => "--flannel-iface={{ flannel_iface }}",
-                                 "kube_vip_tag_version" => "v0.4.2",
-                                 "metal_lb_speaker_tag_version" => "v0.12.1",
-                                 "metal_lb_controller_tag_version" => "v0.12.1",
+                                 "kube_vip_tag_version" => "v0.5.0",
+                                 "metal_lb_speaker_tag_version" => "v0.13.4",
+                                 "metal_lb_controller_tag_version" => "v0.13.4",
                                  "metal_lb_ip_range" => "192.168.30.80-192.168.30.90",
                                  "retry_count" => "30"}
         }

vagrant/test_cluster.py

@@ -0,0 +1,114 @@
+#!/usr/bin/env python3
+
+# Perform a few tests on a cluster created with this playbook.
+# To simplify test execution, the scripts does not depend on any third-party
+# packages, only the Python standard library.
+
+import json
+import subprocess
+import unittest
+from pathlib import Path
+from time import sleep
+from warnings import warn
+
+
+VAGRANT_DIR = Path(__file__).parent.absolute()
+PLAYBOOK_DIR = VAGRANT_DIR.parent.absolute()
+
+
+class TestK3sCluster(unittest.TestCase):
+    def _kubectl(self, args: str, json_out: bool = True) -> dict | None:
+        cmd = "kubectl"
+        if json_out:
+            cmd += " -o json"
+        cmd += f" {args}"
+
+        result = subprocess.run(cmd, capture_output=True, shell=True, check=True)
+
+        if json_out:
+            return json.loads(result.stdout)
+        else:
+            return None
+
+    def _curl(self, url: str) -> str:
+        options = [
+            "--silent",  # no progress info
+            "--show-error",  # ... but errors should still be shown
+            "--fail",  # set exit code on error
+            "--location",  # follow redirects
+        ]
+        cmd = f'curl {" ".join(options)} "{url}"'
+
+        result = subprocess.run(cmd, capture_output=True, shell=True, check=True)
+        output = result.stdout.decode("utf-8")
+        return output
+
+    def _apply_manifest(self, manifest_file: Path) -> dict:
+        apply_result = self._kubectl(
+            f'apply --filename="{manifest_file}" --cascade="background"'
+        )
+        self.addCleanup(
+            lambda: self._kubectl(
+                f'delete --filename="{manifest_file}"',
+                json_out=False,
+            )
+        )
+        return apply_result
+
+    @staticmethod
+    def _retry(function, retries: int = 5, seconds_between_retries=1):
+        for retry in range(1, retries + 1):
+            try:
+                return function()
+            except Exception as exc:
+                if retry < retries:
+                    sleep(seconds_between_retries)
+                    continue
+                else:
+                    raise exc
+
+    def _get_load_balancer_ip(
+        self,
+        service: str,
+        namespace: str = "default",
+    ) -> str | None:
+        svc_description = self._kubectl(
+            f'get --namespace="{namespace}" service "{service}"'
+        )
+        ip = svc_description["status"]["loadBalancer"]["ingress"][0]["ip"]
+        return ip
+
+    def test_nodes_exist(self):
+        out = self._kubectl("get nodes")
+        node_names = {item["metadata"]["name"] for item in out["items"]}
+        self.assertEqual(
+            node_names,
+            {"control1", "control2", "control3", "node1", "node2"},
+        )
+
+    def test_ip_address_pool_exists(self):
+        out = self._kubectl("get --all-namespaces IpAddressPool")
+        pools = out["items"]
+        self.assertGreater(len(pools), 0)
+
+    def test_nginx_example_page(self):
+        # Deploy the manifests to the cluster
+        deployment = self._apply_manifest(PLAYBOOK_DIR / "example" / "deployment.yml")
+        service = self._apply_manifest(PLAYBOOK_DIR / "example" / "service.yml")
+
+        # Assert that the dummy page is available
+        metallb_ip = self._retry(
+            lambda: self._get_load_balancer_ip(service["metadata"]["name"])
+        )
+        # Now that an IP address was assigned, let's reload the service description:
+        service = self._kubectl(f'get service "{service["metadata"]["name"]}"')
+        metallb_port = service["spec"]["ports"][0]["port"]
+
+        response_body = self._retry(
+            lambda: self._curl(f"http://{metallb_ip}:{metallb_port}/")
+        )
+        self.assertIn("Welcome to nginx!", response_body)
+
+
+if __name__ == "__main__":
+    unittest.main()