review change

Author: h3adex

services/ske/wait/wait.go

@@ -3,7 +3,6 @@ package wait
 import (
 	"context"
 	"fmt"
-	"strings"
 	"time"
 
 	"github.com/stackitcloud/stackit-sdk-go/core/wait"
@@ -34,6 +33,8 @@ const (
 	CredentialsRotationStateCompleted = "COMPLETED"
 	// Deprecated: InvalidArgusInstanceErrorCode is deprecated and will be removed after 14th November 2025. Use [ske.RUNTIMEERRORCODE_OBSERVABILITY_INSTANCE_NOT_FOUND] instead.
 	InvalidArgusInstanceErrorCode = "SKE_ARGUS_INSTANCE_NOT_FOUND"
+
+	ClusterNoValidHostFound = "SKE_NODE_NO_VALID_HOST_FOUND"
 )
 
 type APIClientClusterInterface interface {
@@ -43,6 +44,8 @@ type APIClientClusterInterface interface {
 
 // CreateOrUpdateClusterWaitHandler will wait for cluster creation or update
 func CreateOrUpdateClusterWaitHandler(ctx context.Context, a APIClientClusterInterface, projectId, region, name string) *wait.AsyncActionHandler[ske.Cluster] {
+	startTime := time.Now()
+
 	handler := wait.New(func() (waitFinished bool, response *ske.Cluster, err error) {
 		s, err := a.GetClusterExecute(ctx, projectId, region, name)
 		if err != nil {
@@ -57,15 +60,25 @@ func CreateOrUpdateClusterWaitHandler(ctx context.Context, a APIClientClusterInt
 			return true, s, nil
 		}
 
-		if s.Status.Errors != nil && len(*s.Status.Errors) > 0 {
-			var errMsgs []string
+		// If cluster is UNSPECIFIED or UNHEALTHY and has structured errors, exit early
+		if (state == ske.CLUSTERSTATUSSTATE_UNSPECIFIED || state == ske.CLUSTERSTATUSSTATE_UNHEALTHY) &&
+			s.Status.Errors != nil && len(*s.Status.Errors) > 0 {
 			for _, clusterError := range *s.Status.Errors {
-				if clusterError.Message != nil && clusterError.Code != nil {
-					errMsgs = append(errMsgs, fmt.Sprintf("%s %s", *clusterError.Code, *clusterError.Message))
+				if clusterError.Code != nil && clusterError.Message != nil {
+					return true, s, nil
 				}
 			}
+		}
 
-			return true, s, fmt.Errorf("%s", strings.Join(errMsgs, "\n"))
+		// Waiter has been running more than 15 minutes and cluster is still in CREATING or RECONCILING with errors
+		if time.Since(startTime) > 15*time.Minute &&
+			(state == ske.CLUSTERSTATUSSTATE_CREATING || state == ske.CLUSTERSTATUSSTATE_RECONCILING) &&
+			s.Status.Errors != nil && len(*s.Status.Errors) > 0 {
+			for _, clusterError := range *s.Status.Errors {
+				if clusterError.Code != nil && clusterError.Message != nil {
+					return true, s, nil
+				}
+			}
 		}
 
 		if state == ske.CLUSTERSTATUSSTATE_HEALTHY || state == ske.CLUSTERSTATUSSTATE_HIBERNATED {