feat: update function_app_dist-setting workflow to make changes on PR

.github/workflows/push-to-main.yaml → .github/workflows/update-function-app-dist.yaml

@@ -1,9 +1,10 @@
-name: Push to main branch
+name: Update function_app_dist
 
 on:
-  push:
+  pull_request:
     branches:
       - main
+      - dev
 
 jobs:
   update_function_app_dist:
@@ -16,8 +17,13 @@ jobs:
         uses: actions/checkout@v2
 
       - name: Update function_app_dist default to 'release'
+        if: github.event.pull_request.base.ref == 'main'
         run: sed -i 's/^\( *default =\) "dev"/\1 "release"/' variables.tf 
 
+      - name: Update function_app_dist default to 'dev'
+        if: github.event.pull_request.base.ref == 'dev'
+        run: sed -i 's/^\( *default =\) "release"/\1 "dev"/' variables.tf 
+
       - name: Commit changes
         uses: stefanzweifel/git-auto-commit-action@v4
         with:

blob.tf

@@ -1,4 +1,5 @@
 resource "azurerm_storage_account" "deployment_sa" {
+  count                    = var.deployment_storage_account_name == "" ? 1 : 0
   name                     = "${local.alphanumeric_prefix_name}${local.alphanumeric_cluster_name}deployment"
   location                 = data.azurerm_resource_group.rg.location
   resource_group_name      = var.rg_name
@@ -12,16 +13,17 @@ resource "azurerm_storage_account" "deployment_sa" {
 }
 
 resource "azurerm_storage_container" "deployment" {
+  count                 = var.deployment_container_name == "" ? 1 : 0
   name                  = "${local.alphanumeric_prefix_name}${local.alphanumeric_cluster_name}-deployment"
-  storage_account_name  = azurerm_storage_account.deployment_sa.name
+  storage_account_name  = local.deployment_storage_account_name
   container_access_type = "private"
   depends_on            = [azurerm_storage_account.deployment_sa]
 }
 
 resource "azurerm_storage_blob" "state" {
   name                   = "state"
-  storage_account_name   = azurerm_storage_account.deployment_sa.name
-  storage_container_name = azurerm_storage_container.deployment.name
+  storage_account_name   = local.deployment_storage_account_name
+  storage_container_name = local.deployment_container_name
   type                   = "Block"
   source_content         = "{\"initial_size\":${var.cluster_size}, \"desired_size\":${var.cluster_size}, \"instances\":[], \"clusterized\":false}"
   depends_on             = [azurerm_storage_container.deployment]
@@ -30,3 +32,15 @@ resource "azurerm_storage_blob" "state" {
     ignore_changes = all
   }
 }
+
+data azurerm_storage_account "deployment_blob" {
+  count               = var.deployment_storage_account_name != "" ? 1 : 0
+  name                = var.deployment_storage_account_name
+  resource_group_name = var.rg_name
+}
+
+data "azurerm_storage_account" "obs_sa" {
+  count               = var.obs_name != "" ? 1 : 0
+  name                = var.obs_name
+  resource_group_name = var.rg_name
+}

clients.tf

@@ -0,0 +1,49 @@
+resource "null_resource" "get-backend-ip" {
+  count = var.clients_number > 0 ? 1 : 0
+  triggers = {
+    always_run = timestamp()
+  }
+  provisioner "local-exec" {
+    command = "az vmss nic list -g ${var.rg_name} --vmss-name ${azurerm_linux_virtual_machine_scale_set.vmss.name} --subscription ${var.subscription_id} --query \"[].ipConfigurations[]\" | jq -r '.[] | select(.name==\"ipconfig0\")'.privateIPAddress > ${path.root}/backend_ips"
+  }
+  depends_on = [azurerm_linux_virtual_machine_scale_set.vmss]
+}
+
+data "local_file" "backend_ips" {
+  count      = var.clients_number > 0 ? 1 : 0
+  filename   = "${path.root}/backend_ips"
+  depends_on = [null_resource.get-backend-ip]
+}
+
+module "clients" {
+  count              = var.clients_number > 0 ? 1 : 0
+  source             = "./modules/clients"
+  rg_name            = var.rg_name
+  clients_name       = "${var.prefix}-${var.cluster_name}-client"
+  clients_number     = var.clients_number
+  mount_clients_dpdk = var.mount_clients_dpdk
+  subnet_name        = var.subnet_name
+  apt_repo_url       = var.apt_repo_url
+  vnet_name          = var.vnet_name
+  nics               = var.mount_clients_dpdk ? var.client_nics_num : 1
+  instance_type      = var.client_instance_type
+  backend_ips        = [replace(join(" ",[data.local_file.backend_ips[0].content]), "\n", " ")]
+  ssh_public_key     = var.ssh_public_key == null ? tls_private_key.ssh_key[0].public_key_openssh : var.ssh_public_key
+  ppg_id             = local.placement_group_id
+  assign_public_ip   = var.private_network ? false : true
+  vnet_rg_name       = var.vnet_rg_name
+  source_image_id    = var.source_image_id
+  sg_id              = var.sg_id
+  depends_on         = [azurerm_proximity_placement_group.ppg]
+}
+
+resource "null_resource" "clean" {
+  count = var.clients_number > 0 ? 1 : 0
+  triggers = {
+    always_run = timestamp()
+  }
+  provisioner "local-exec" {
+    command = "rm -f ${path.root}/backend_ips"
+  }
+  depends_on = [module.clients]
+}

examples/existing_private_network/outputs.tf

@@ -1,3 +1,3 @@
 output "get-cluster-helpers-commands" {
-  value = module.deploy-weka.cluster_helpers_commands
+  value = module.deploy-weka
 }

examples/existing_private_network_with_peering/outputs.tf

@@ -1,3 +1,3 @@
 output "get-cluster-helpers-commands" {
-  value = module.deploy-weka.cluster_helpers_commands
+  value = module.deploy-weka
 }

examples/public_network/README.md

@@ -33,6 +33,7 @@ No resources.
 | <a name="input_prefix"></a> [prefix](#input\_prefix) | Prefix for all resources | `string` | n/a | yes |
 | <a name="input_rg_name"></a> [rg\_name](#input\_rg\_name) | Name of existing resource group | `string` | n/a | yes |
 | <a name="input_set_obs_integration"></a> [set\_obs\_integration](#input\_set\_obs\_integration) | Should be true to enable OBS integration with weka cluster | `bool` | n/a | yes |
+| <a name="input_sg_ssh_range"></a> [sg\_ssh\_range](#input\_sg\_ssh\_range) | A list of IP addresses that can use ssh connection with a public network deployment. | `list(string)` | `[]` | no |
 | <a name="input_subnet_delegation"></a> [subnet\_delegation](#input\_subnet\_delegation) | Subnet delegation enables you to designate a specific subnet for an Azure PaaS service | `string` | n/a | yes |
 | <a name="input_subnet_prefixes"></a> [subnet\_prefixes](#input\_subnet\_prefixes) | Address prefixes to use for the subnet | `string` | n/a | yes |
 | <a name="input_subscription_id"></a> [subscription\_id](#input\_subscription\_id) | Subscription id for deployment | `string` | n/a | yes |

examples/public_network/main.tf

@@ -11,6 +11,7 @@ module "create-network" {
   rg_name           = var.rg_name
   address_space     = var.address_space
   subnet_prefixes   = var.subnet_prefixes
+  sg_ssh_range      = var.sg_ssh_range
 }
 
 module "deploy-weka" {

examples/public_network/outputs.tf

@@ -1,3 +1,3 @@
 output "get-cluster-helpers-commands" {
-  value = module.deploy-weka.cluster_helpers_commands
+  value = module.deploy-weka
 }

examples/public_network/variables.tf

@@ -65,4 +65,10 @@ variable "tiering_ssd_percent" {
 variable "cluster_size" {
   type = number
   description = "Weka cluster size"
-}
+}
+
+variable "sg_ssh_range" {
+  type        = list(string)
+  description = "A list of IP addresses that can use ssh connection with a public network deployment."
+  default = []
+}

examples/public_network/vars.auto.tfvars

@@ -7,4 +7,5 @@ cluster_name        = "poc"
 instance_type       = "Standard_L8s_v3"
 set_obs_integration = true
 tiering_ssd_percent = 20
-cluster_size        = 6
+cluster_size        = 6
+sg_ssh_range        = ["0.0.0.0/0"]

examples/public_network_with_existing_obs/README.md

@@ -36,6 +36,7 @@ No resources.
 | <a name="input_prefix"></a> [prefix](#input\_prefix) | Prefix for all resources | `string` | n/a | yes |
 | <a name="input_rg_name"></a> [rg\_name](#input\_rg\_name) | Name of existing resource group | `string` | n/a | yes |
 | <a name="input_set_obs_integration"></a> [set\_obs\_integration](#input\_set\_obs\_integration) | Should be true to enable OBS integration with weka cluster | `bool` | n/a | yes |
+| <a name="input_sg_ssh_range"></a> [sg\_ssh\_range](#input\_sg\_ssh\_range) | A list of IP addresses that can use ssh connection with a public network deployment. | `list(string)` | `[]` | no |
 | <a name="input_subnet_delegation"></a> [subnet\_delegation](#input\_subnet\_delegation) | Subnet delegation enables you to designate a specific subnet for an Azure PaaS service | `string` | n/a | yes |
 | <a name="input_subnet_prefixes"></a> [subnet\_prefixes](#input\_subnet\_prefixes) | Address prefixes to use for the subnet | `string` | n/a | yes |
 | <a name="input_subscription_id"></a> [subscription\_id](#input\_subscription\_id) | Subscription id for deployment | `string` | n/a | yes |

examples/public_network_with_existing_obs/main.tf

@@ -11,6 +11,7 @@ module "create-network" {
   rg_name           = var.rg_name
   address_space     = var.address_space
   subnet_prefixes   = var.subnet_prefixes
+  sg_ssh_range      = var.sg_ssh_range
 }
 
 module "deploy-weka" {

examples/public_network_with_existing_obs/outputs.tf

@@ -1,3 +1,3 @@
 output "get-cluster-helpers-commands" {
-  value = module.deploy-weka.cluster_helpers_commands
+  value = module.deploy-weka
 }

examples/public_network_with_existing_obs/variables.tf

@@ -81,4 +81,10 @@ variable "blob_obs_access_key" {
   type = string
   description = "Access key of existing blob obs container"
   sensitive   = true
-}
+}
+
+variable "sg_ssh_range" {
+  type        = list(string)
+  description = "A list of IP addresses that can use ssh connection with a public network deployment."
+  default = []
+}

examples/public_network_with_existing_obs/vars.auto.tfvars

@@ -10,4 +10,5 @@ obs_name            = "obs-blob"
 obs_container_name  = "obs-container"
 blob_obs_access_key = ""
 tiering_ssd_percent = 20
-cluster_size        = 6
+cluster_size        = 6
+sg_ssh_range        = ["0.0.0.0/0"]

function-app/code/common/common.go

@@ -21,6 +21,7 @@ import (
 	"github.com/weka/go-cloud-lib/lib/types"
 	"github.com/weka/go-cloud-lib/logging"
 	"github.com/weka/go-cloud-lib/protocol"
+	reportLib "github.com/weka/go-cloud-lib/report"
 )
 
 type InvokeRequest struct {
@@ -34,14 +35,13 @@ type InvokeResponse struct {
 	ReturnValue interface{}
 }
 
-type ClusterState struct {
-	InitialSize int                 `json:"initial_size"`
-	DesiredSize int                 `json:"desired_size"`
-	Progress    map[string][]string `json:"progress"`
-	Errors      map[string][]string `json:"errors"`
-	Instances   []string            `json:"instances"`
-	Clusterized bool                `json:"clusterized"`
-}
+const FindDrivesScript = `
+import json
+import sys
+for d in json.load(sys.stdin)['disks']:
+	if d['isRotational']: continue
+	print(d['devPath'])
+`
 
 func leaseContainer(ctx context.Context, subscriptionId, resourceGroupName, storageAccountName, containerName string, leaseIdIn *string, action armstorage.LeaseContainerRequestAction) (leaseIdOut *string, err error) {
 	logger := logging.LoggerFromCtx(ctx)
@@ -129,7 +129,7 @@ func ReadBlobObject(ctx context.Context, stateStorageName, containerName, blobNa
 
 }
 
-func ReadState(ctx context.Context, stateStorageName, containerName string) (state ClusterState, err error) {
+func ReadState(ctx context.Context, stateStorageName, containerName string) (state protocol.ClusterState, err error) {
 	logger := logging.LoggerFromCtx(ctx)
 
 	stateAsByteArray, err := ReadBlobObject(ctx, stateStorageName, containerName, "state")
@@ -166,7 +166,7 @@ func WriteBlobObject(ctx context.Context, stateStorageName, containerName, blobN
 
 }
 
-func WriteState(ctx context.Context, stateStorageName, containerName string, state ClusterState) (err error) {
+func WriteState(ctx context.Context, stateStorageName, containerName string, state protocol.ClusterState) (err error) {
 	logger := logging.LoggerFromCtx(ctx)
 
 	stateAsByteArray, err := json.Marshal(state)
@@ -191,7 +191,7 @@ func (e *ShutdownRequired) Error() string {
 	return e.Message
 }
 
-func AddInstanceToState(ctx context.Context, subscriptionId, resourceGroupName, stateStorageName, stateContainerName, newInstance string) (state ClusterState, err error) {
+func AddInstanceToState(ctx context.Context, subscriptionId, resourceGroupName, stateStorageName, stateContainerName, newInstance string) (state protocol.ClusterState, err error) {
 	logger := logging.LoggerFromCtx(ctx)
 
 	leaseId, err := LockContainer(ctx, subscriptionId, resourceGroupName, stateStorageName, stateContainerName)
@@ -229,7 +229,7 @@ func AddInstanceToState(ctx context.Context, subscriptionId, resourceGroupName,
 	return
 }
 
-func UpdateClusterized(ctx context.Context, subscriptionId, resourceGroupName, stateStorageName, stateContainerName string) (state ClusterState, err error) {
+func UpdateClusterized(ctx context.Context, subscriptionId, resourceGroupName, stateStorageName, stateContainerName string) (state protocol.ClusterState, err error) {
 	logger := logging.LoggerFromCtx(ctx)
 
 	leaseId, err := LockContainer(ctx, subscriptionId, resourceGroupName, stateStorageName, stateContainerName)
@@ -835,12 +835,19 @@ func RetrySetDeletionProtectionAndReport(
 
 		if protectionErr, ok := err.(*azcore.ResponseError); ok && protectionErr.ErrorCode == "AuthorizationFailed" {
 			counter++
+			// deletion protection invoked by terminate function
+			if maxAttempts == 0 {
+				msg := fmt.Sprintf("Deletion protection set authorization isn't ready, will retry on next scale down workflow")
+				ReportMsg(ctx, hostName, subscriptionId, resourceGroupName, stateContainerName, stateStorageName, "debug", msg)
+				return
+			}
+
 			if counter > maxAttempts {
 				break
 			}
-			msg := fmt.Sprintf("Setting deletion protection authorization error, going to sleep for %dM", sleepInterval)
+			msg := fmt.Sprintf("Deletion protection set authorization isn't ready, going to sleep for %s", sleepInterval)
 			logger.Info().Msg(msg)
-			ReportMsg(ctx, hostName, subscriptionId, resourceGroupName, stateContainerName, stateStorageName, "progress", msg)
+			ReportMsg(ctx, hostName, subscriptionId, resourceGroupName, stateContainerName, stateStorageName, "debug", msg)
 			time.Sleep(sleepInterval)
 		} else {
 			break
@@ -854,7 +861,7 @@ func RetrySetDeletionProtectionAndReport(
 }
 
 func ReportMsg(ctx context.Context, hostName, subscriptionId, resourceGroupName, stateContainerName, stateStorageName, reportType, message string) {
-	reportObj := Report{Type: reportType, Hostname: hostName, Message: message}
+	reportObj := protocol.Report{Type: reportType, Hostname: hostName, Message: message}
 	_ = UpdateStateReporting(ctx, subscriptionId, resourceGroupName, stateContainerName, stateStorageName, reportObj)
 }
 
@@ -944,13 +951,7 @@ func TerminateScaleSetInstances(ctx context.Context, subscriptionId, resourceGro
 	return
 }
 
-type Report struct {
-	Type     string `json:"type"`
-	Message  string `json:"message"`
-	Hostname string `json:"hostname"`
-}
-
-func UpdateStateReporting(ctx context.Context, subscriptionId, resourceGroupName, stateContainerName, stateStorageName string, report Report) (err error) {
+func UpdateStateReporting(ctx context.Context, subscriptionId, resourceGroupName, stateContainerName, stateStorageName string, report protocol.Report) (err error) {
 	logger := logging.LoggerFromCtx(ctx)
 
 	leaseId, err := LockContainer(ctx, subscriptionId, resourceGroupName, stateStorageName, stateContainerName)
@@ -970,31 +971,19 @@ func UpdateStateReporting(ctx context.Context, subscriptionId, resourceGroupName
 	return
 }
 
-func UpdateStateReportingWithoutLocking(ctx context.Context, stateContainerName, stateStorageName string, report Report) (err error) {
+func UpdateStateReportingWithoutLocking(ctx context.Context, stateContainerName, stateStorageName string, report protocol.Report) (err error) {
 	state, err := ReadState(ctx, stateStorageName, stateContainerName)
 	if err != nil {
 		return
 	}
-	currentTime := time.Now().UTC().Format("15:04:05") + " UTC"
-	switch report.Type {
-	case "error":
-		if state.Errors == nil {
-			state.Errors = make(map[string][]string)
-		}
-		state.Errors[report.Hostname] = append(state.Errors[report.Hostname], fmt.Sprintf("%s: %s", currentTime, report.Message))
-	case "progress":
-		if state.Progress == nil {
-			state.Progress = make(map[string][]string)
-		}
-		state.Progress[report.Hostname] = append(state.Progress[report.Hostname], fmt.Sprintf("%s: %s", currentTime, report.Message))
-	default:
-		err = fmt.Errorf("invalid type: %s", report.Type)
+	err = reportLib.UpdateReport(report, &state)
+	if err != nil {
+		err = fmt.Errorf("failed updating state report")
 		return
 	}
-
 	err = WriteState(ctx, stateStorageName, stateContainerName, state)
 	if err != nil {
-		err = fmt.Errorf("failed updating state errors")
+		err = fmt.Errorf("failed updating state report")
 		return
 	}
 	return