[PTX-1641] Added AKS node driver

deployments/deploy-ssh.sh

@@ -85,6 +85,21 @@ if [ -z "$STORAGENODE_RECOVERY_TIMEOUT" ]; then
     echo "Using default storage node recovery timeout of ${STORAGENODE_RECOVERY_TIMEOUT}"
 fi
 
+AZURE_TENANTID=""
+if [ -n "$AZURE_TENANT_ID" ]; then
+    AZURE_TENANTID="${AZURE_TENANT_ID}"
+fi
+
+AZURE_CLIENTID=""
+if [ -n "$AZURE_CLIENT_ID" ]; then
+    AZURE_CLIENTID="${AZURE_CLIENT_ID}"
+fi
+
+AZURE_CLIENTSECRET=""
+if [ -n "$AZURE_CLIENT_SECRET" ]; then
+    AZURE_CLIENTSECRET="${AZURE_CLIENT_SECRET}"
+fi
+
 if [ -z "$TEST_SUITE" ]; then
     TEST_SUITE='"bin/asg.test",
             "bin/autopilot-capacity.test",
@@ -183,6 +198,7 @@ if [ -n "${K8S_VENDOR}" ]; then
             K8S_VENDOR_KEY=px/enabled
             K8S_VENDOR_OPERATOR="In"
             K8S_VENDOR_VALUE='values: ["false"]'
+            NODE_DRIVER="aks"
             ;;
         eks)
             # Run torpedo on worker node, where px installation is disabled.
@@ -306,6 +322,12 @@ spec:
       value: "${TORPEDO_SSH_PASSWORD}"
     - name: TORPEDO_SSH_KEY
       value: "${TORPEDO_SSH_KEY}"
+    - name: AZURE_TENANT_ID
+      value: "${AZURE_TENANTID}"
+    - name: AZURE_CLIENT_ID
+      value: "${AZURE_CLIENTID}"
+    - name: AZURE_CLIENT_SECRET
+      value: "${AZURE_CLIENTSECRET}"
   volumes: [${VOLUMES}]
   restartPolicy: Never
   serviceAccountName: torpedo-account
@@ -337,4 +359,4 @@ for i in $(seq 1 600) ; do
 done
 
 echo "Error: Failed to wait for torpedo to start running..."
-describe_pod_then_exit
+describe_pod_then_exit

drivers/node/aks/aks.go

@@ -0,0 +1,80 @@
+package aks
+
+import (
+	"os"
+	"time"
+
+	"github.com/libopenstorage/cloudops"
+	"github.com/libopenstorage/cloudops/azure"
+	"github.com/portworx/torpedo/drivers/node"
+	"github.com/portworx/torpedo/drivers/node/ssh"
+	"github.com/sirupsen/logrus"
+)
+
+const (
+	// DriverName is the name of the gke driver
+	DriverName = "aks"
+	// ZoneCount number of zones in autoscaling group
+	ZoneCount = 3
+)
+
+type aks struct {
+	ssh.SSH
+	ops           cloudops.Ops
+	instanceGroup string
+}
+
+func (a *aks) String() string {
+	return DriverName
+}
+
+func (a *aks) Init() error {
+
+	instanceGroup := os.Getenv("INSTANCE_GROUP")
+	if len(instanceGroup) != 0 {
+		a.instanceGroup = instanceGroup
+	} else {
+		a.instanceGroup = "nodepool1"
+	}
+
+	ops, err := azure.NewClientFromMetadata()
+	if err != nil {
+		return err
+	}
+	a.ops = ops
+
+	return nil
+}
+
+func (a *aks) SetASGClusterSize(perZoneCount int64, timeout time.Duration) error {
+	// Azure SDK requires total cluster size
+	totalClusterSize := perZoneCount * ZoneCount
+	err := a.ops.SetInstanceGroupSize(a.instanceGroup, totalClusterSize, timeout)
+	if err != nil {
+		logrus.Errorf("failed to set size of node pool %s. Error: %v", a.instanceGroup, err)
+		return err
+	}
+
+	return nil
+}
+
+func (a *aks) GetASGClusterSize() (int64, error) {
+	nodeCount, err := a.ops.GetInstanceGroupSize(a.instanceGroup)
+	if err != nil {
+		logrus.Errorf("failed to get size of node pool %s. Error: %v", a.instanceGroup, err)
+		return 0, err
+	}
+
+	return nodeCount, nil
+}
+
+func init() {
+
+	SSHDriver := ssh.SSH{}
+	SSHDriver.Init()
+	g := &aks{
+		SSH: SSHDriver,
+	}
+
+	node.Register(DriverName, g)
+}

drivers/node/gke/gke.go

@@ -45,8 +45,9 @@ func (g *gke) Init() error {
 	return nil
 }
 
-func (g *gke) SetASGClusterSize(count int64, timeout time.Duration) error {
-	err := g.ops.SetInstanceGroupSize(g.instanceGroup, count, timeout)
+func (g *gke) SetASGClusterSize(perZoneCount int64, timeout time.Duration) error {
+	// GCP SDK requires per zone cluster size
+	err := g.ops.SetInstanceGroupSize(g.instanceGroup, perZoneCount, timeout)
 	if err != nil {
 		logrus.Errorf("failed to set size of node pool %s. Error: %v", g.instanceGroup, err)
 		return err