Tets PR for Integration test

.github/workflows/basic-ci.yaml

@@ -0,0 +1,98 @@
+name: Harvester-Network-Controller CI
+
+# synchronize for pull request update
+on:
+  push:
+    branches:
+      - master
+      - 'v**'
+  pull_request:
+    types: [opened, reopened, synchronize]
+
+env:
+  LIBVIRT_DEFAULT_URI: "qemu:///system"
+  VM_DEPLOYED: false
+
+jobs:
+  validation:
+    runs-on:
+      - self-hosted
+      - golang
+    steps:
+      - name: "Clone and check"
+        uses: actions/checkout@v3
+      - name: "Run validations"
+        run: |
+          make validate
+  job-new-installation:
+    needs: validation
+    runs-on:
+      - self-hosted
+      - golang
+    steps:
+      - name: "Clone and check"
+        uses: actions/checkout@v3
+      - name: "Build the Image for the Integration Test"
+        run: |
+          BUILD_FOR_CI=true make
+          ./ci/scripts/patch-ttl-repo.sh
+          echo "NCH override result as below:"
+          cat ci/charts/nch-override.yaml
+      - name: "Local Deployment (Harvester+NetworkController) for testing"
+        id: vm_deploy
+        run: |
+          rm -rf nch-new-vagrant-k3s
+          git clone https://github.com/bk201/vagrant-k3s nch-new-vagrant-k3s
+          pushd nch-new-vagrant-k3s
+          yq e -i ".cluster_size = 1" settings.yaml
+          ./new-cluster.sh
+          echo "VM_DEPLOYED=true" >> "$GITHUB_ENV"
+          yq e -i ".longhorn_version = \"1.7.1\"" settings.yaml
+          ./scripts/deploy_longhorn.sh
+          popd 
+      - name: "Patch Image target"
+        run: |
+          ./ci/scripts/patch-ttl-repo.sh
+          echo "NCH override result as below:"
+          cat ci/charts/nch-override.yaml
+      - name: "Deploy NCH"
+        run: |
+          pushd nch-new-vagrant-k3s
+          cp ../ci/scripts/deploy_nch_current.sh ./deploy_nch_current.sh
+          cp ../ci/charts/nch-override.yaml ./nch-override.yaml
+          ./deploy_nch_current.sh
+          popd
+      - name: "Add disk"
+        run: |
+          pushd nch-new-vagrant-k3s
+          ./scripts/attach-disk.sh node1 nch-new-vagrant-k3s
+          sleep 30
+          popd
+      - name: "Run Basic Test"
+        id: basic-test
+        run: |
+          pushd nch-new-vagrant-k3s
+          vagrant ssh-config node1 > ../ssh-config
+          cp kubeconfig ../kubeconfig
+          popd
+          echo Running integration tests
+          NCH_HOME=`pwd` go test -v ./tests/...
+      - name: "Get NCH logs"
+        if: always()
+        run: |
+          if [ ${{ env.VM_DEPLOYED }} != 'true' ]; then
+            echo "VM is not deployed, skip getting logs"
+            exit 0
+          fi
+          ./ci/scripts/get-debug-info.sh
+      - name: "Tear Down / Cleanup"
+        if: always()
+        run: |
+          if [ ${{ env.VM_DEPLOYED }} != 'true' ]; then
+            echo "VM is not deployed, skip VM destroy"
+            exit 0
+          fi
+          rm -rf /tmp/network-controller/nch-new-vagrant-k3s
+          pushd nch-new-vagrant-k3s
+          vagrant destroy -f --parallel
+          popd

ci/charts/nch-override.yaml

@@ -0,0 +1,17 @@
+image:
+  repository: rancher/harvester-network-controller
+  pullPolicy: IfNotPresent
+  # Overrides the image tag whose default is the chart appVersion.
+  tag: ""
+
+webhook:
+  repository: rancher/harvester-network-webhook
+  pullPolicy: IfNotPresent
+  # Overrides the image tag whose default is the chart appVersion.
+  tag: ""
+
+autoProvisionFilter: [/dev/sd*]
+debug: true
+
+# we only manually inject udev monitor error once, so we can test it in CI.
+injectUdevMonitorError: true

ci/scripts/deploy_nch_chart.sh

@@ -0,0 +1,68 @@
+#!/bin/bash -e
+
+TOP_DIR="$( cd "$( dirname "${BASH_SOURCE[0]}" )/" &> /dev/null && pwd )"
+
+ensure_command() {
+  local cmd=$1
+  if ! which $cmd &> /dev/null; then
+    echo 1
+    return
+  fi
+  echo 0
+}
+
+ensure_network_controller_ready() {
+  while [ true ]; do
+    running_num=$(kubectl get pods -n harvester-system |grep ^network-controller |grep Running |awk '{print $3}' |wc -l)
+    if [[ $running_num -eq ${cluster_nodes} ]]; then
+      echo "network-controller pods are ready!"
+      break
+    fi
+    echo "check network-controller pods failure."
+    exit 1
+  done
+}
+
+if [ ! -f $TOP_DIR/kubeconfig ]; then
+  echo "kubeconfig does not exist. Please create cluster first."
+  echo "Maybe try new_cluster.sh"
+  exit 1
+fi
+echo $TOP_DIR/kubeconfig
+export KUBECONFIG=$TOP_DIR/kubeconfig
+
+if [[ $(ensure_command helm) -eq 1 ]]; then
+  echo "no helm, try to curl..."
+  curl -O https://get.helm.sh/helm-v3.9.4-linux-amd64.tar.gz
+  tar -zxvf helm-v3.9.4-linux-amd64.tar.gz
+  HELM=$TOP_DIR/linux-amd64/helm
+  $HELM version
+else
+  echo "Get helm, version info as below"
+  HELM=$(which helm)
+  $HELM version
+fi
+
+pushd $TOP_DIR
+
+cat >> nch-override.yaml.default << 'EOF'
+autoProvisionFilter: [/dev/sd*]
+EOF
+
+if [ ! -f nch-override.yaml ]; then
+  mv nch-override.yaml.default nch-override.yaml
+fi
+
+#$HELM pull harvester-network-controller --repo https://charts.harvesterhci.io --untar
+#$HELM install -f $TOP_DIR/nch-override.yaml harvester-network-controller ./harvester-network-controller --create-namespace -n harvester-system
+
+$HELM repo add harvester https://charts.harvesterhci.io
+$HELM repo update
+kubectl apply -f https://raw.githubusercontent.com/harvester/network-controller-harvester/master/manifests/dependency_crds
+$HELM install harvester-network-controller harvester/harvester-network-controller
+
+cluster_nodes=$(yq -e e '.cluster_size' $TOP_DIR/settings.yaml)
+echo "cluster nodes: $cluster_nodes"
+ensure_network_controller_ready
+
+popd

ci/scripts/deploy_nch_current.sh

@@ -0,0 +1,65 @@
+#!/bin/bash -e
+
+TOP_DIR="$( cd "$( dirname "${BASH_SOURCE[0]}" )/" &> /dev/null && pwd )"
+
+ensure_command() {
+  local cmd=$1
+  if ! which $cmd &> /dev/null; then
+    echo 1
+    return
+  fi
+  echo 0
+}
+
+ensure_network_controller_ready() {
+  while [ true ]; do
+    running_num=$(kubectl get pods -n harvester-system |grep ^network-controller |grep Running |awk '{print $3}' |wc -l)
+    if [[ $running_num -eq ${cluster_nodes} ]]; then
+      echo "network-controller pods are ready!"
+      break
+    fi
+    echo "check network-controller failure."
+    exit 1
+  done
+}
+
+if [ ! -f $TOP_DIR/kubeconfig ]; then
+  echo "kubeconfig does not exist. Please create cluster first."
+  echo "Maybe try new_cluster.sh"
+  exit 1
+fi
+echo $TOP_DIR/kubeconfig
+export KUBECONFIG=$TOP_DIR/kubeconfig
+
+if [[ $(ensure_command helm) -eq 1 ]]; then
+  echo "no helm, try to curl..."
+  curl -O https://get.helm.sh/helm-v3.9.4-linux-amd64.tar.gz
+  tar -zxvf helm-v3.9.4-linux-amd64.tar.gz
+  HELM=$TOP_DIR/linux-amd64/helm
+  $HELM version
+else
+  echo "Get helm, version info as below"
+  HELM=$(which helm)
+  $HELM version
+fi
+
+cluster_nodes=$(yq -e e '.cluster_size' $TOP_DIR/settings.yaml)
+echo "cluster nodes: $cluster_nodes"
+ensure_network_controller_ready
+
+pushd $TOP_DIR
+cat >> nch-override.yaml.default << 'EOF'
+autoProvisionFilter: [/dev/sd*]
+EOF
+
+if [ ! -f nch-override.yaml ]; then
+  mv nch-override.yaml.default nch-override.yaml
+fi
+
+cp -r ../deploy/charts/harvester-node-disk-manager harvester-network-cntroller
+
+target_img=$(yq -e .image.repository nch-override.yaml)
+echo "install target image: ${target_img}"
+$HELM install -f $TOP_DIR/nch-override.yaml harvester-network-cntroller ./harvester-network-cntroller --create-namespace -n harvester-system
+
+popd

ci/scripts/get-debug-info.sh

@@ -0,0 +1,10 @@
+#!/bin/bash -e
+
+TARGETNODE="node1"
+
+export KUBECONFIG=kubeconfig
+
+NDMPOD=$(kubectl get pods -n harvester-system --field-selector spec.nodeName=$TARGETNODE | grep ^harvester-network-cntroller |grep -v webhook |awk '{print $1}')
+
+# filter out the redundant Skip log
+kubectl logs $NDMPOD -n harvester-system |grep -v Skip

ci/scripts/patch-ttl-repo.sh

@@ -0,0 +1,10 @@
+#!/bin/bash -e
+
+COMMIT=$(git rev-parse --short HEAD)
+IMAGE=ttl.sh/network-controller-harvester-${COMMIT}
+IMAGE_WEBHOOK=ttl.sh/node-disk-manager-webhook-${COMMIT}
+
+yq e -i ".image.repository = \"${IMAGE}\"" ci/charts/nch-override.yaml
+yq e -i ".image.tag = \"1h\"" ci/charts/nch-override.yaml 
+yq e -i ".webhook.image.repository = \"${IMAGE_WEBHOOK}\"" ci/charts/nch-override.yaml
+yq e -i ".webhook.image.tag = \"1h\"" ci/charts/nch-override.yaml 

ci/scripts/upgrade_nch.sh

@@ -0,0 +1,62 @@
+#!/bin/bash -e
+
+TOP_DIR="$( cd "$( dirname "${BASH_SOURCE[0]}" )/" &> /dev/null && pwd )"
+
+ensure_command() {
+  local cmd=$1
+  if ! which $cmd &> /dev/null; then
+    echo 1
+    return
+  fi
+  echo 0
+}
+
+if [ ! -f $TOP_DIR/kubeconfig ]; then
+  echo "kubeconfig does not exist. Please create cluster first."
+  echo "Maybe try new_cluster.sh"
+  exit 1
+fi
+echo $TOP_DIR/kubeconfig
+export KUBECONFIG=$TOP_DIR/kubeconfig
+
+if [[ $(ensure_command helm) -eq 1 ]]; then
+  echo "no helm, try to curl..."
+  curl -O https://get.helm.sh/helm-v3.9.4-linux-amd64.tar.gz
+  tar -zxvf helm-v3.9.4-linux-amd64.tar.gz
+  HELM=$TOP_DIR/linux-amd64/helm
+  $HELM version
+else
+  echo "Get helm, version info as below"
+  HELM=$(which helm)
+  $HELM version
+fi
+
+cluster_nodes=$(yq -e e '.cluster_size' $TOP_DIR/settings.yaml)
+echo "cluster nodes: $cluster_nodes"
+
+pushd $TOP_DIR
+# cleanup first
+rm -rf harvester-node-disk-manager*
+rm -rf ndm-override.yaml
+
+cp -r ../deploy/charts/harvester-node-disk-manager harvester-node-disk-manager
+cp ../ci/charts/ndm-override.yaml ndm-override.yaml
+
+target_img=$(yq -e .image.repository ndm-override.yaml)
+echo "upgrade target image: ${target_img}, upgrading ..."
+$HELM upgrade -f $TOP_DIR/ndm-override.yaml harvester-node-disk-manager harvester-node-disk-manager/ -n harvester-system
+
+sleep 30 # wait 30 seconds for ndm respawn pods
+
+# check image
+pod_name=$(kubectl get pods -n harvester-system |grep Running |grep -v webhook |grep ^harvester-node-disk-manager|head -n1 |awk '{print $1}')
+container_img=$(kubectl get pods ${pod_name} -n harvester-system -o yaml |yq -e .spec.containers[0].image |tr ":" \n)
+yaml_img=$(yq -e .image.repository ndm-override.yaml)
+if grep -q ${yaml_img} <<< ${container_img}; then
+  echo "Image is equal: ${yaml_img}"
+else
+  echo "Image is non-equal, container: ${container_img}, yaml file: ${yaml_img}"
+  exit 1
+fi
+echo "harvester-node-disk-manager upgrade successfully!"
+popd