Refactor ARM64 e2e test suite

- Migrate the GitHub runners to Equinix c3.large.arm64 instances
- Run the test suite on all supported Kubernetes versions
- Add multi-tenancy smoke test

Signed-off-by: Stefan Prodan <stefan.prodan@gmail.com>

.github/runners/README.md

@@ -1,24 +1,32 @@
 # Flux ARM64 GitHub runners
 
-The Flux ARM64 end-to-end tests run on Equinix instances provisioned with Docker and GitHub self-hosted runners.
+The Flux ARM64 end-to-end tests run on Equinix Metal instances provisioned with Docker and GitHub self-hosted runners.
 
 ## Current instances
 
-| Runner        | Instance            | Region |
-|---------------|---------------------|--------|
-| equinix-arm-1 | flux-equinix-arm-01 | AMS1   |
-| equinix-arm-2 | flux-equinix-arm-01 | AMS1   |
-| equinix-arm-3 | flux-equinix-arm-01 | AMS1   |
-| equinix-arm-4 | flux-equinix-arm-02 | DFW2   |
-| equinix-arm-5 | flux-equinix-arm-02 | DFW2   |
-| equinix-arm-6 | flux-equinix-arm-02 | DFW2   |
+| Repository                  | Runner           | Instance               | Location      |
+|-----------------------------|------------------|------------------------|---------------|
+| flux2                       | equinix-arm-dc-1 | flux-equinix-arm-dc-01 | Washington DC |
+| flux2                       | equinix-arm-dc-2 | flux-equinix-arm-dc-01 | Washington DC |
+| flux2                       | equinix-arm-da-1 | flux-equinix-arm-da-01 | Dallas        |
+| flux2                       | equinix-arm-da-2 | flux-equinix-arm-da-01 | Dallas        |
+| source-controller           | equinix-arm-dc-1 | flux-equinix-arm-dc-01 | Washington DC |
+| source-controller           | equinix-arm-da-1 | flux-equinix-arm-da-01 | Dallas        |
+| image-automation-controller | equinix-arm-dc-1 | flux-equinix-arm-dc-01 | Washington DC |
+| image-automation-controller | equinix-arm-da-1 | flux-equinix-arm-da-01 | Dallas        |
+
+Instance spec:
+- Ampere Altra Q80-30 80-core processor @ 2.8GHz
+- 2 x 960GB NVME
+- 256GB RAM
+- 2 x 25Gbps
 
 ## Instance setup
 
 In order to add a new runner to the GitHub Actions pool,
 first create a server on Equinix with the following configuration:
-- Type: c2.large.arm
-- OS: Ubuntu 20.04
+- Type: `c3.large.arm64`
+- OS: `Ubuntu 22.04 LTS`
 
 ### Install prerequisites
 
@@ -54,14 +62,14 @@ sudo ./prereq.sh
 
 - Retrieve the GitHub runner token from the repository [settings page](https://github.com/fluxcd/flux2/settings/actions/runners/new?arch=arm64&os=linux)
 
-- Create 3 directories `runner1`, `runner2`, `runner3`
+- Create two directories `flux2-01`, `flux2-02`
 
 - In each dir run:
 ```shell
 curl -sL https://raw.githubusercontent.com/fluxcd/flux2/main/.github/runners/runner-setup.sh > runner-setup.sh \
   && chmod +x ./runner-setup.sh
 
-./runner-setup.sh equinix-arm-<NUMBER> <TOKEN>
+./runner-setup.sh equinix-arm-<NUMBER> <TOKEN> <REPO>
 ```
 
 - Reboot the instance

.github/runners/prereq.sh

@@ -18,11 +18,11 @@
 
 set -eu
 
-KIND_VERSION=0.14.0
+KIND_VERSION=0.17.0
 KUBECTL_VERSION=1.24.0
-KUSTOMIZE_VERSION=4.5.4
-HELM_VERSION=3.8.2
-GITHUB_RUNNER_VERSION=2.291.1
+KUSTOMIZE_VERSION=4.5.7
+HELM_VERSION=3.10.1
+GITHUB_RUNNER_VERSION=2.298.2
 PACKAGES="apt-transport-https ca-certificates software-properties-common build-essential libssl-dev gnupg lsb-release jq pkg-config"
 
 # install prerequisites
@@ -31,6 +31,10 @@ apt-get update \
   && apt-get clean \
   && rm -rf /var/lib/apt/lists/*
 
+# fix Kubernetes DNS resolution
+rm /etc/resolv.conf
+cat "/run/systemd/resolve/stub-resolv.conf" | sed '/search/d' > /etc/resolv.conf
+
 # install docker
 curl -fsSL https://get.docker.com -o get-docker.sh \
   && chmod +x get-docker.sh

.github/runners/runner-setup.sh

@@ -22,7 +22,7 @@ RUNNER_NAME=$1
 REPOSITORY_TOKEN=$2
 REPOSITORY_URL=${3:-https://github.com/fluxcd/flux2}
 
-GITHUB_RUNNER_VERSION=2.285.1
+GITHUB_RUNNER_VERSION=2.298.2
 
 # download runner
 curl -o actions-runner-linux-arm64.tar.gz -L https://github.com/actions/runner/releases/download/v${GITHUB_RUNNER_VERSION}/actions-runner-linux-arm64-${GITHUB_RUNNER_VERSION}.tar.gz \

.github/workflows/e2e-arm64.yaml

@@ -3,7 +3,7 @@ name: e2e-arm64
 on:
   workflow_dispatch:
   push:
-    branches: [ main, update-components ]
+    branches: [ main, update-components, e2e-arm64* ]
 
 permissions:
   contents: read
@@ -13,6 +13,10 @@ jobs:
     # Hosted on Equinix
     # Docs: https://github.com/fluxcd/flux2/tree/main/.github/runners
     runs-on: [self-hosted, Linux, ARM64, equinix]
+    strategy:
+      matrix:
+        # Keep this list up-to-date with https://endoflife.date/kubernetes
+        KUBERNETES_VERSION: [ 1.23.13, 1.24.7, 1.25.3 ]
     steps:
       - name: Checkout
         uses: actions/checkout@v3
@@ -23,16 +27,46 @@ jobs:
       - name: Prepare
         id: prep
         run: |
-          echo ::set-output name=CLUSTER::arm64-${GITHUB_SHA:0:7}-$(date +%s)
-          echo ::set-output name=CONTEXT::kind-arm64-${GITHUB_SHA:0:7}-$(date +%s)
+          ID=${GITHUB_SHA:0:7}-${{ matrix.KUBERNETES_VERSION }}-$(date +%s)
+          echo "CLUSTER=arm64-${ID}" >> $GITHUB_OUTPUT
       - name: Build
         run: |
           make build
       - name: Setup Kubernetes Kind
         run: |
-          kind create cluster --name ${{ steps.prep.outputs.CLUSTER }} --kubeconfig=/tmp/${{ steps.prep.outputs.CLUSTER }}
+          kind create cluster \
+          --wait 5m \
+          --name ${{ steps.prep.outputs.CLUSTER }} \
+          --kubeconfig=/tmp/${{ steps.prep.outputs.CLUSTER }} \
+          --image=kindest/node:v${{ matrix.KUBERNETES_VERSION }}
       - name: Run e2e tests
         run: TEST_KUBECONFIG=/tmp/${{ steps.prep.outputs.CLUSTER }} make e2e
+      - name: Run multi-tenancy tests
+        env:
+          KUBECONFIG: /tmp/${{ steps.prep.outputs.CLUSTER }}
+        run: |
+          ./bin/flux install
+          ./bin/flux create source git flux-system \
+          --interval=15m \
+          --url=https://github.com/fluxcd/flux2-multi-tenancy \
+          --branch=main \
+          --ignore-paths="./clusters/**/flux-system/"
+          ./bin/flux create kustomization flux-system \
+          --interval=15m \
+          --source=flux-system \
+          --path=./clusters/staging
+          kubectl -n flux-system wait kustomization/tenants --for=condition=ready --timeout=5m
+          kubectl -n apps wait kustomization/dev-team --for=condition=ready --timeout=1m
+          kubectl -n apps wait helmrelease/podinfo --for=condition=ready --timeout=1m
+      - name: Debug failure
+        if: failure()
+        env:
+          KUBECONFIG: /tmp/${{ steps.prep.outputs.CLUSTER }}
+        run: |
+          kubectl -n flux-system get all
+          kubectl -n flux-system describe po 
+          kubectl -n flux-system logs deploy/source-controller
+          kubectl -n flux-system logs deploy/kustomize-controller
       - name: Cleanup
         if: always()
         run: |