Merge pull request #617 from rostachen/master

tencentcloud_kubernetes_cluster add upgrade_instances_follow_cluster

Author: rostachen

CHANGELOG.md

@@ -1,4 +1,10 @@
-## 1.56.3 (Unreleased)
+## 1.56.4 (Unreleased)
+## 1.56.3 (April 25, 2021)
+
+ENHANCEMENTS:
+
+* Resource: `tencentcloud_kubernetes_cluster` add `upgrade_instances_follow_cluster` for upgrade all instances of cluster.
+
 ## 1.56.2 (April 19, 2021)
 
 BUG FIXES:

tencentcloud/resource_tc_kubernetes_cluster.go

@@ -661,6 +661,12 @@ func resourceTencentCloudTkeCluster() *schema.Resource {
 			Default:     "1.10.5",
 			Description: "Version of the cluster, Default is '1.10.5'.",
 		},
+		"upgrade_instances_follow_cluster": {
+			Type:        schema.TypeBool,
+			Optional:    true,
+			Default:     false,
+			Description: "Indicates whether upgrade all instances when cluster_version change. Default is false.",
+		},
 		"cluster_ipvs": {
 			Type:        schema.TypeBool,
 			ForceNew:    true,
@@ -673,7 +679,7 @@ func resourceTencentCloudTkeCluster() *schema.Resource {
 			ForceNew:    true,
 			Optional:    true,
 			Default:     false,
-			Description: "Indicates whether to enable cluster node auto scaler.",
+			Description: "Indicates whether to enable cluster node auto scaler. Default is false.",
 		},
 		"node_pool_global_config": {
 			Type:     schema.TypeList,
@@ -979,11 +985,11 @@ func resourceTencentCloudTkeCluster() *schema.Resource {
 			Description: "Labels of tke cluster nodes.",
 		},
 		"unschedulable": {
-			Type:        schema.TypeInt,
-			Optional:    true,
-			ForceNew:    true,
+			Type:     schema.TypeInt,
+			Optional: true,
+			ForceNew: true,
 			DiffSuppressFunc: func(k, old, new string, d *schema.ResourceData) bool {
-				if new == "0" && old == ""{
+				if new == "0" && old == "" {
 					return true
 				} else {
 					return old == new
@@ -1299,6 +1305,57 @@ func tkeGetNodePoolGlobalConfig(d *schema.ResourceData) *tke.ModifyClusterAsGrou
 	return request
 }
 
+// upgradeClusterInstances upgrade instances, upgrade type try seq:major, hot.
+func upgradeClusterInstances(tkeService TkeService, ctx context.Context, id string) error {
+	// get all available instances for upgrade
+	upgradeType := "major"
+	instanceIds, err := tkeService.CheckInstancesUpgradeAble(ctx, id, upgradeType)
+	if err != nil {
+		return err
+	}
+	if len(instanceIds) == 0 {
+		upgradeType = "hot"
+		instanceIds, err = tkeService.CheckInstancesUpgradeAble(ctx, id, upgradeType)
+		if err != nil {
+			return err
+		}
+	}
+	log.Println("instancesIds for upgrade:", instanceIds)
+	if len(instanceIds) == 0 {
+		return nil
+	}
+
+	// upgrade instances
+	err = resource.Retry(writeRetryTimeout, func() *resource.RetryError {
+		inErr := tkeService.UpgradeClusterInstances(ctx, id, upgradeType, instanceIds)
+		if inErr != nil {
+			return retryError(inErr)
+		}
+		return nil
+	})
+	if err != nil {
+		return err
+	}
+
+	// check update status
+	err = resource.Retry(readRetryTimeout, func() *resource.RetryError {
+		done, inErr := tkeService.GetUpgradeInstanceResult(ctx, id)
+		if inErr != nil {
+			return retryError(inErr)
+		}
+		if done {
+			return nil
+		} else {
+			return resource.RetryableError(fmt.Errorf("cluster %s, retry...", id))
+		}
+	})
+	if err != nil {
+		return err
+	}
+
+	return nil
+}
+
 func resourceTencentCloudTkeClusterCreate(d *schema.ResourceData, meta interface{}) error {
 	defer logElapsed("resource.tencentcloud_kubernetes_cluster.create")()
 
@@ -2114,7 +2171,7 @@ func resourceTencentCloudTkeClusterUpdate(d *schema.ResourceData, meta interface
 		d.SetPartial("cluster_desc")
 	}
 
-	//upgrate k8s version
+	//upgrade k8s cluster version
 	if d.HasChange("cluster_version") {
 		newVersion := d.Get("cluster_version").(string)
 		isOk, err := tkeService.CheckClusterVersion(ctx, id, newVersion)
@@ -2156,6 +2213,18 @@ func resourceTencentCloudTkeClusterUpdate(d *schema.ResourceData, meta interface
 		if err != nil {
 			return err
 		}
+
+		// upgrade instances version
+		upgrade := false
+		if v, ok := d.GetOk("upgrade_instances_follow_cluster"); ok {
+			upgrade = v.(bool)
+		}
+		if upgrade {
+			err := upgradeClusterInstances(tkeService, ctx, id)
+			if err != nil {
+				return err
+			}
+		}
 	}
 
 	// update node pool global config

tencentcloud/service_tencentcloud_tke.go

@@ -328,6 +328,61 @@ func (me *TkeService) DescribeClusterConfig(ctx context.Context, id string) (con
 	return
 }
 
+func (me *TkeService) GetUpgradeInstanceResult(ctx context.Context, id string) (
+	done bool,
+	errRet error,
+) {
+
+	logId := getLogId(ctx)
+	request := tke.NewGetUpgradeInstanceProgressRequest()
+
+	defer func() {
+		if errRet != nil {
+			log.Printf("[CRITAL]%s api[%s] fail, request body [%s], reason[%s]\n",
+				logId, request.GetAction(), request.ToJsonString(), errRet.Error())
+		}
+	}()
+
+	request.ClusterId = &id
+
+	ratelimit.Check(request.GetAction())
+	response, err := me.client.UseTkeClient().GetUpgradeInstanceProgress(request)
+
+	if err != nil {
+		errRet = err
+		return
+	}
+
+	lifeState := *response.Response.LifeState
+
+	// all instances success, lifeState=done
+	if lifeState == "done" {
+		return true, nil
+	} else if lifeState != "process" {
+		return false, fmt.Errorf("upgrade instances failed, tke response lifeState is:%s", lifeState)
+	}
+
+	// parent lifeState=process, check whether all instances in processing.
+	for _, inst := range response.Response.Instances {
+		if *inst.LifeState == "done" || *inst.LifeState == "pending" {
+			continue
+		}
+		if *inst.LifeState != "process" {
+			return false, fmt.Errorf("upgrade instances failed, " +
+				"instanceId:%s, lifeState is:%s", *inst.InstanceID, *inst.LifeState)
+		}
+		// instance lifeState=process, check whether failed or not.
+		for _, detail := range inst.Detail {
+			if *detail.LifeState == "failed" {
+				return false, fmt.Errorf("upgrade instances failed, " +
+					"instanceId:%s, detail.lifeState is:%s", *inst.InstanceID, *detail.LifeState)
+			}
+		}
+	}
+
+	return
+}
+
 func (me *TkeService) CreateCluster(ctx context.Context,
 	basic ClusterBasicSetting,
 	advanced ClusterAdvancedSettings,
@@ -915,6 +970,58 @@ func (me *TkeService) CheckClusterVersion(ctx context.Context, id string, cluste
 	return
 }
 
+func (me *TkeService) CheckInstancesUpgradeAble(ctx context.Context, id string, upgradeType string) (instanceIds []string, errRet error) {
+	logId := getLogId(ctx)
+	request := tke.NewCheckInstancesUpgradeAbleRequest()
+	defer func() {
+		if errRet != nil {
+			log.Printf("[CRITAL]%s api[%s] fail, reason[%s]\n", logId, request.GetAction(), errRet.Error())
+		}
+	}()
+	request.ClusterId = &id
+	request.UpgradeType = &upgradeType
+	ratelimit.Check(request.GetAction())
+
+	resp, err := me.client.UseTkeClient().CheckInstancesUpgradeAble(request)
+	if err != nil {
+		errRet = err
+		return
+	}
+
+	if resp == nil || resp.Response == nil || resp.Response.UpgradeAbleInstances == nil {
+		return
+	}
+	for _, inst := range resp.Response.UpgradeAbleInstances {
+		instanceIds = append(instanceIds, *inst.InstanceId)
+	}
+
+	return
+}
+
+func (me *TkeService) UpgradeClusterInstances(ctx context.Context, id string, upgradeType string, instanceIds []string) (errRet error) {
+	logId := getLogId(ctx)
+	request := tke.NewUpgradeClusterInstancesRequest()
+	defer func() {
+		if errRet != nil {
+			log.Printf("[CRITAL]%s api[%s] fail, reason[%s]\n", logId, request.GetAction(), errRet.Error())
+		}
+	}()
+	op := "create"
+	request.Operation = &op
+	request.ClusterId = &id
+	request.UpgradeType = &upgradeType
+	request.InstanceIds = helper.Strings(instanceIds)
+	ratelimit.Check(request.GetAction())
+
+	_, err := me.client.UseTkeClient().UpgradeClusterInstances(request)
+	if err != nil {
+		errRet = err
+		return
+	}
+
+	return
+}
+
 func (me *TkeService) DescribeImages(ctx context.Context) (imageIds []string, errRet error) {
 	logId := getLogId(ctx)
 	request := tke.NewDescribeImagesRequest()