feat: refactor scale_up, consider vmss refresh mode in other funcs

README.md

@@ -408,7 +408,7 @@ proxy_url = VALUE
 | <a name="input_function_app_storage_account_prefix"></a> [function\_app\_storage\_account\_prefix](#input\_function\_app\_storage\_account\_prefix) | Weka storage account name prefix | `string` | `"weka"` | no |
 | <a name="input_function_app_subnet_delegation_cidr"></a> [function\_app\_subnet\_delegation\_cidr](#input\_function\_app\_subnet\_delegation\_cidr) | Subnet delegation enables you to designate a specific subnet for an Azure PaaS service. | `string` | `"10.0.1.0/25"` | no |
 | <a name="input_function_app_subnet_delegation_id"></a> [function\_app\_subnet\_delegation\_id](#input\_function\_app\_subnet\_delegation\_id) | Required to specify if subnet\_name were used to specify pre-defined subnets for weka. Function subnet delegation requires an additional subnet, and in the case of pre-defined networking this one also should be pre-defined | `string` | `""` | no |
-| <a name="input_function_app_version"></a> [function\_app\_version](#input\_function\_app\_version) | Function app code version (hash) | `string` | `"60ec8c21cc5fd2b21d698fa9bb9e69ac"` | no |
+| <a name="input_function_app_version"></a> [function\_app\_version](#input\_function\_app\_version) | Function app code version (hash) | `string` | `"93a261d4128c2c3aa7809fecc4b94746"` | no |
 | <a name="input_get_weka_io_token"></a> [get\_weka\_io\_token](#input\_get\_weka\_io\_token) | The token to download the Weka release from get.weka.io. | `string` | `""` | no |
 | <a name="input_hotspare"></a> [hotspare](#input\_hotspare) | Number of hotspares to set on weka cluster. Refer to https://docs.weka.io/overview/ssd-capacity-management#hot-spare | `number` | `1` | no |
 | <a name="input_install_cluster_dpdk"></a> [install\_cluster\_dpdk](#input\_install\_cluster\_dpdk) | Install weka cluster with DPDK | `bool` | `true` | no |
@@ -459,6 +459,7 @@ proxy_url = VALUE
 | <a name="input_tiering_obs_name"></a> [tiering\_obs\_name](#input\_tiering\_obs\_name) | Name of existing obs storage account | `string` | `""` | no |
 | <a name="input_traces_per_ionode"></a> [traces\_per\_ionode](#input\_traces\_per\_ionode) | The number of traces per ionode. Traces are low-level events generated by Weka processes and are used as troubleshooting information for support purposes. | `number` | `10` | no |
 | <a name="input_vm_username"></a> [vm\_username](#input\_vm\_username) | Provided as part of output for automated use of terraform, in case of custom AMI and automated use of outputs replace this with user that should be used for ssh connection | `string` | `"weka"` | no |
+| <a name="input_vmss_instances_adding_step"></a> [vmss\_instances\_adding\_step](#input\_vmss\_instances\_adding\_step) | Number of instances to add to vmss in one iteration | `number` | `3` | no |
 | <a name="input_vmss_single_placement_group"></a> [vmss\_single\_placement\_group](#input\_vmss\_single\_placement\_group) | Sets single\_placement\_group option for vmss. If true, a scale set is composed of a single placement group, and has a range of 0-100 VMs. | `bool` | `true` | no |
 | <a name="input_vnet_name"></a> [vnet\_name](#input\_vnet\_name) | The virtual network name. | `string` | `""` | no |
 | <a name="input_vnet_rg_name"></a> [vnet\_rg\_name](#input\_vnet\_rg\_name) | Resource group name of vnet. Will be used when vnet\_name is not provided. | `string` | `""` | no |

blob.tf

@@ -52,10 +52,9 @@ resource "azurerm_storage_blob" "vmss_state" {
   type                   = "Block"
 
   source_content = jsonencode({
-    vmss_created   = false
     vmss_version   = 0
-    refresh_status = 0
-    current_config = {}
+    refresh_status = "none"
+    current_config = null
   })
 
   lifecycle {

function-app/code/common/common.go

@@ -748,6 +748,17 @@ func getScaleSet(ctx context.Context, subscriptionId, resourceGroupName, vmScale
 	return &scaleSet.VirtualMachineScaleSet, nil
 }
 
+func GetScaleSetOrNilOnNotFound(ctx context.Context, subscriptionId, resourceGroupName, vmScaleSetName string) (*armcompute.VirtualMachineScaleSet, error) {
+	vmss, err := getScaleSet(ctx, subscriptionId, resourceGroupName, vmScaleSetName)
+	if err != nil {
+		if getErr, ok := err.(*azcore.ResponseError); ok && getErr.ErrorCode == "ResourceNotFound" {
+			return nil, nil
+		}
+		return nil, err
+	}
+	return vmss, nil
+}
+
 // Gets single scale set info
 func GetScaleSetInfo(ctx context.Context, subscriptionId, resourceGroupName, vmScaleSetName, keyVaultUri string) (*ScaleSetInfo, error) {
 	logger := logging.LoggerFromCtx(ctx)

function-app/code/common/vmss_config.go

@@ -95,20 +95,15 @@ func VmssConfigsDiff(old, new *VMSSConfig) string {
 	return cmp.Diff(new, old) // arguments order: (want, got)
 }
 
-type RefreshStatus uint8
+type RefreshStatus string
 
 const (
-	RefreshNone       RefreshStatus = iota
-	RefreshInProgress RefreshStatus = iota
-	RefreshNeeded     RefreshStatus = iota
+	RefreshNone       RefreshStatus = "none"
+	RefreshInProgress RefreshStatus = "in_progress"
+	RefreshNeeded     RefreshStatus = "needed"
 )
 
-func (s *RefreshStatus) String() string {
-	return []string{"none", "in_progress", "needed"}[*s]
-}
-
 type VMSSState struct {
-	VmssCreated   bool          `json:"vmss_created"`
 	VmssVersion   uint16        `json:"vmss_version"`
 	RefreshStatus RefreshStatus `json:"refresh_status"`
 	CurrentConfig *VMSSConfig   `json:"current_config,omitempty"`

function-app/code/functions/clusterize/clusterize.go

@@ -58,7 +58,6 @@ type ClusterizationParams struct {
 	Location          string
 	Prefix            string
 	KeyVaultUri       string
-	VmssVersion       uint16
 
 	StateContainerName string
 	StateStorageName   string
@@ -97,7 +96,8 @@ func HandleLastClusterVm(ctx context.Context, state protocol.ClusterState, p Clu
 	logger := logging.LoggerFromCtx(ctx)
 	logger.Info().Msg("This is the last instance in the cluster, creating obs and clusterization script")
 
-	vmScaleSetName := common.GetVmScaleSetName(p.Prefix, p.Cluster.ClusterName, p.VmssVersion)
+	version := 0 // on cluserization step we are sure that the vmss version is 0 as no refresh was done yet
+	vmScaleSetName := common.GetVmScaleSetName(p.Prefix, p.Cluster.ClusterName, uint16(version))
 
 	if p.Cluster.SetObs {
 		if p.Obs.AccessKey == "" {
@@ -178,7 +178,8 @@ func Clusterize(ctx context.Context, p ClusterizationParams) (clusterizeScript s
 
 	instanceName := strings.Split(p.VmName, ":")[0]
 	instanceId := common.GetScaleSetVmIndex(instanceName)
-	vmScaleSetName := common.GetVmScaleSetName(p.Prefix, p.Cluster.ClusterName, p.VmssVersion)
+	version := 0 // on cluserization step we are sure that the vmss version is 0 as no refresh was done yet
+	vmScaleSetName := common.GetVmScaleSetName(p.Prefix, p.Cluster.ClusterName, uint16(version))
 	vmName := p.VmName
 
 	ip, err := common.GetPublicIp(ctx, p.SubscriptionId, p.ResourceGroupName, vmScaleSetName, p.Prefix, p.Cluster.ClusterName, instanceId)
@@ -227,7 +228,6 @@ func Clusterize(ctx context.Context, p ClusterizationParams) (clusterizeScript s
 func Handler(w http.ResponseWriter, r *http.Request) {
 	stateContainerName := os.Getenv("STATE_CONTAINER_NAME")
 	stateStorageName := os.Getenv("STATE_STORAGE_NAME")
-	vmssStateStorageName := os.Getenv("VMSS_STATE_STORAGE_NAME")
 	hostsNum, _ := strconv.Atoi(os.Getenv("HOSTS_NUM"))
 	clusterName := os.Getenv("CLUSTER_NAME")
 	subscriptionId := os.Getenv("SUBSCRIPTION_ID")
@@ -284,21 +284,12 @@ func Handler(w http.ResponseWriter, r *http.Request) {
 		return
 	}
 
-	vmssState, err := common.ReadVmssState(ctx, vmssStateStorageName, stateContainerName)
-	if err != nil {
-		err = fmt.Errorf("cannot read vmss state to read get vmss version: %v", err)
-		logger.Error().Err(err).Send()
-		common.WriteErrorResponse(w, err)
-		return
-	}
-
 	params := ClusterizationParams{
 		SubscriptionId:     subscriptionId,
 		ResourceGroupName:  resourceGroupName,
 		Location:           location,
 		Prefix:             prefix,
 		KeyVaultUri:        keyVaultUri,
-		VmssVersion:        vmssState.VmssVersion,
 		StateContainerName: stateContainerName,
 		StateStorageName:   stateStorageName,
 		VmName:             data.Vm,

function-app/code/functions/join_finalization/join_finalization.go

@@ -60,6 +60,10 @@ func Handler(w http.ResponseWriter, r *http.Request) {
 	}
 
 	vmScaleSetName := common.GetVmScaleSetName(prefix, clusterName, vmssState.VmssVersion)
+	// use the refresh vmss name if the cluster is in refresh mode
+	if vmssState.RefreshStatus != common.RefreshNone {
+		vmScaleSetName = common.GetRefreshVmssName(vmScaleSetName, vmssState.VmssVersion)
+	}
 
 	err = common.SetDeletionProtection(ctx, subscriptionId, resourceGroupName, vmScaleSetName, common.GetScaleSetVmIndex(data.Name), true)
 	if err != nil {

function-app/code/functions/protect/protect.go

@@ -63,6 +63,10 @@ func Handler(w http.ResponseWriter, r *http.Request) {
 	}
 
 	vmScaleSetName := common.GetVmScaleSetName(prefix, clusterName, vmssState.VmssVersion)
+	// use the refresh vmss name if the cluster is in refresh mode
+	if vmssState.RefreshStatus != common.RefreshNone {
+		vmScaleSetName = common.GetRefreshVmssName(vmScaleSetName, vmssState.VmssVersion)
+	}
 
 	instanceName := strings.Split(data.Vm, ":")[0]
 	hostName := strings.Split(data.Vm, ":")[1]

function-app/code/functions/resize/resize.go

@@ -77,6 +77,11 @@ func Handler(w http.ResponseWriter, r *http.Request) {
 	}
 
 	vmssName := common.GetVmScaleSetName(prefix, clusterName, vmssState.VmssVersion)
+	// use the refresh vmss name if the cluster is in refresh mode
+	if vmssState.RefreshStatus != common.RefreshNone {
+		vmssName = common.GetRefreshVmssName(vmssName, vmssState.VmssVersion)
+	}
+
 	err = updateDesiredClusterSize(ctx, *size.Value, subscriptionId, resourceGroupName, vmssName, stateContainerName, stateStorageName)
 	if err != nil {
 		logger.Error().Err(err).Send()