fix the bug of create cluster crash and cluster_os force replacement when upgrade

Author: ivan

CHANGELOG.md

@@ -1,4 +1,11 @@
-## 1.56.11 (Unreleased)
+## 1.56.12 (Unreleased)
+
+## 1.56.11 
+
+BUG FIXES
+ 
+* Resource `tencentcloud_kubernetes_cluster` fix create cluster without *desired_pod_num* in tf, then crash
+* Resource `tencentcloud_kubernetes_cluster` fix when upgrade terraform-provider-tencentclod from v1.56.1 to newer，cluster_os force replacement
 
 ## 1.56.10 
 

tencentcloud/extension_tke.go

@@ -1,5 +1,7 @@
 package tencentcloud
 
+import tke "github.com/tencentcloud/tencentcloud-sdk-go/tencentcloud/tke/v20180525"
+
 const (
 	TKE_CLUSTER_OS_CENTOS72 = "centos7.2x86_64"
 	TKE_CLUSTER_OS_CENTOS76 = "centos7.6.0_x64"
@@ -9,6 +11,13 @@ const (
 	TKE_CLUSTER_OS_LINUXF22 = "Tencent tlinux release 2.2 (Final)"
 )
 
+// 兼容旧的 cluster_os 定义
+const (
+	TkeClusterOsCentOS76 = "centos7.6x86_64"
+	//TkeClusterOsUbuntu16 = "ubuntu16.04.1 LTSx86_64"
+	TkeClusterOsUbuntu18 = "ubuntu18.04.1 LTSx86_64"
+)
+
 var TKE_CLUSTER_OS = []string{TKE_CLUSTER_OS_CENTOS76, TKE_CLUSTER_OS_UBUNTU18, TKE_CLUSTER_OS_LINUX24}
 
 var tkeClusterOsMap = map[string]string{TKE_CLUSTER_OS_CENTOS72: TKE_CLUSTER_OS_CENTOS72,
@@ -85,3 +94,8 @@ const (
 )
 
 var TKE_CLUSTER_KUBE_PROXY_MODE = []string{TKE_CLUSTER_KUBE_PROXY_MODE_BPF}
+
+type OverrideSettings struct {
+	Master []tke.InstanceAdvancedSettings
+	Work   []tke.InstanceAdvancedSettings
+}

tencentcloud/resource_tc_kubernetes_cluster.go

@@ -1623,7 +1623,10 @@ func resourceTencentCloudTkeClusterCreate(d *schema.ResourceData, meta interface
 		}
 	}
 
-	overrideSettings := make([]tke.InstanceAdvancedSettings, 0)
+	overrideSettings := OverrideSettings{
+		Master: make([]tke.InstanceAdvancedSettings, 0),
+		Work:   make([]tke.InstanceAdvancedSettings, 0),
+	}
 	if masters, ok := d.GetOk("master_config"); ok {
 		if clusterDeployType == TKE_DEPLOY_TYPE_MANAGED {
 			return fmt.Errorf("if `cluster_deploy_type` is `MANAGED_CLUSTER` , You don't need define the master yourself")
@@ -1643,7 +1646,7 @@ func resourceTencentCloudTkeClusterCreate(d *schema.ResourceData, meta interface
 			if v, ok := master["desired_pod_num"]; ok {
 				dpNum := int64(v.(int))
 				if dpNum != -1 {
-					overrideSettings = append(overrideSettings, tke.InstanceAdvancedSettings{DesiredPodNumber: helper.Int64(dpNum)})
+					overrideSettings.Master = append(overrideSettings.Master, tke.InstanceAdvancedSettings{DesiredPodNumber: helper.Int64(dpNum)})
 				}
 			}
 		}
@@ -1670,7 +1673,7 @@ func resourceTencentCloudTkeClusterCreate(d *schema.ResourceData, meta interface
 			if v, ok := worker["desired_pod_num"]; ok {
 				dpNum := int64(v.(int))
 				if dpNum != -1 {
-					overrideSettings = append(overrideSettings, tke.InstanceAdvancedSettings{DesiredPodNumber: helper.Int64(dpNum)})
+					overrideSettings.Work = append(overrideSettings.Work, tke.InstanceAdvancedSettings{DesiredPodNumber: helper.Int64(dpNum)})
 				}
 			}
 		}
@@ -1716,7 +1719,7 @@ func resourceTencentCloudTkeClusterCreate(d *schema.ResourceData, meta interface
 	}
 
 	service := TkeService{client: meta.(*TencentCloudClient).apiV3Conn}
-	id, err := service.CreateCluster(ctx, basic, advanced, cvms, iAdvanced, cidrSet, tags, existIntances, overrideSettings)
+	id, err := service.CreateCluster(ctx, basic, advanced, cvms, iAdvanced, cidrSet, tags, existIntances, &overrideSettings)
 	if err != nil {
 		return err
 	}
@@ -1893,9 +1896,18 @@ func resourceTencentCloudTkeClusterRead(d *schema.ResourceData, meta interface{}
 		return nil
 	}
 
+	// 兼容旧的 cluster_os 的 key, 由于 cluster_os有默认值，所以不大可能为空
+	oldOs := d.Get("cluster_os").(string)
+	newOs := tkeToShowClusterOs(info.ClusterOs)
+
+	if (oldOs == TkeClusterOsCentOS76 && newOs == TKE_CLUSTER_OS_CENTOS76) ||
+		(oldOs == TkeClusterOsUbuntu18 && newOs == TKE_CLUSTER_OS_UBUNTU18) {
+		newOs = oldOs
+	}
+
 	_ = d.Set("cluster_name", info.ClusterName)
 	_ = d.Set("cluster_desc", info.ClusterDescription)
-	_ = d.Set("cluster_os", tkeToShowClusterOs(info.ClusterOs))
+	_ = d.Set("cluster_os", newOs)
 	_ = d.Set("cluster_deploy_type", info.DeployType)
 	_ = d.Set("cluster_version", info.ClusterVersion)
 	_ = d.Set("cluster_ipvs", info.Ipvs)

tencentcloud/service_tencentcloud_tke.go

@@ -31,7 +31,7 @@ type ClusterBasicSetting struct {
 type ClusterAdvancedSettings struct {
 	Ipvs                    bool
 	AsEnabled               bool
-	EnableCustomizedPodCIDR	bool
+	EnableCustomizedPodCIDR bool
 	BasePodNumber           int64
 	ContainerRuntime        string
 	NodeNameType            string
@@ -371,13 +371,13 @@ func (me *TkeService) GetUpgradeInstanceResult(ctx context.Context, id string) (
 			continue
 		}
 		if *inst.LifeState != "process" {
-			return false, fmt.Errorf("upgrade instances failed, " +
+			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, " +
+				return false, fmt.Errorf("upgrade instances failed, "+
 					"instanceId:%s, detail.lifeState is:%s", *inst.InstanceID, *detail.LifeState)
 			}
 		}
@@ -394,7 +394,7 @@ func (me *TkeService) CreateCluster(ctx context.Context,
 	cidrSetting ClusterCidrSettings,
 	tags map[string]string,
 	existedInstance []*tke.ExistedInstancesForNode,
-	overrideSettings []tke.InstanceAdvancedSettings,
+	overrideSettings *OverrideSettings,
 ) (id string, errRet error) {
 
 	logId := getLogId(ctx)
@@ -459,7 +459,8 @@ func (me *TkeService) CreateCluster(ctx context.Context,
 		request.InstanceAdvancedSettings.Labels = iAdvanced.Labels
 	}
 
-	if len(overrideSettings) > 0 && len(overrideSettings) != (len(cvms.Master)+len(cvms.Work)) {
+	if len(overrideSettings.Master)+len(overrideSettings.Work) > 0 &&
+		len(overrideSettings.Master)+len(overrideSettings.Work) != (len(cvms.Master)+len(cvms.Work)) {
 		return "", fmt.Errorf("len(overrideSettings) != (len(cvms.Master)+len(cvms.Work))")
 	}
 
@@ -473,7 +474,9 @@ func (me *TkeService) CreateCluster(ctx context.Context,
 		request.ClusterType = helper.String(TKE_DEPLOY_TYPE_INDEPENDENT)
 		for v := range cvms.Master {
 			node.RunInstancesPara = append(node.RunInstancesPara, &cvms.Master[v])
-			node.InstanceAdvancedSettingsOverrides = append(node.InstanceAdvancedSettingsOverrides, &overrideSettings[v])
+			if len(overrideSettings.Master) != 0 {
+				node.InstanceAdvancedSettingsOverrides = append(node.InstanceAdvancedSettingsOverrides, &overrideSettings.Master[v])
+			}
 		}
 		request.RunInstancesForNode = append(request.RunInstancesForNode, &node)
 
@@ -487,7 +490,9 @@ func (me *TkeService) CreateCluster(ctx context.Context,
 		node.RunInstancesPara = []*string{}
 		for v := range cvms.Work {
 			node.RunInstancesPara = append(node.RunInstancesPara, &cvms.Work[v])
-			node.InstanceAdvancedSettingsOverrides = append(node.InstanceAdvancedSettingsOverrides, &overrideSettings[v])
+			if len(overrideSettings.Work) != 0 {
+				node.InstanceAdvancedSettingsOverrides = append(node.InstanceAdvancedSettingsOverrides, &overrideSettings.Work[v])
+			}
 		}
 		request.RunInstancesForNode = append(request.RunInstancesForNode, &node)
 	}
@@ -940,9 +945,9 @@ func (me *TkeService) ModifyClusterVersion(ctx context.Context, id string, clust
 		}
 
 		request.ExtraArgs = &tke.ClusterExtraArgs{
-			KubeAPIServer: kas,
+			KubeAPIServer:         kas,
 			KubeControllerManager: kcms,
-			KubeScheduler: kss,
+			KubeScheduler:         kss,
 		}
 	}