From df11df2616c32d091adec707a20421f3a7c0aab0 Mon Sep 17 00:00:00 2001
From: stefanSpectro <stefan.knott@spectrocloud.com>
Date: Thu, 22 Feb 2024 16:03:06 -0700
Subject: [PATCH] return set of issues when ng is in degraded state

---
 pkg/cloud/services/eks/nodegroup.go      | 26 +++++++++++++
 pkg/cloud/services/eks/nodegroup_test.go | 48 ++++++++++++++++++++++++
 2 files changed, 74 insertions(+)
 create mode 100644 pkg/cloud/services/eks/nodegroup_test.go

diff --git a/pkg/cloud/services/eks/nodegroup.go b/pkg/cloud/services/eks/nodegroup.go
index ec24f16fc3..acffe3cdd3 100644
--- a/pkg/cloud/services/eks/nodegroup.go
+++ b/pkg/cloud/services/eks/nodegroup.go
@@ -19,6 +19,8 @@ package eks
 import (
 	"context"
 	"fmt"
+	"slices"
+	"strings"
 
 	"github.com/aws/aws-sdk-go/aws"
 	"github.com/aws/aws-sdk-go/aws/awserr"
@@ -37,6 +39,7 @@ import (
 	"sigs.k8s.io/cluster-api-provider-aws/v2/pkg/cloud/services/wait"
 	"sigs.k8s.io/cluster-api-provider-aws/v2/pkg/record"
 	clusterv1 "sigs.k8s.io/cluster-api/api/v1beta1"
+	capierrors "sigs.k8s.io/cluster-api/errors"
 	"sigs.k8s.io/cluster-api/util/annotations"
 )
 
@@ -588,6 +591,29 @@ func (s *NodegroupService) setStatus(ng *eks.Nodegroup) error {
 		managedPool.Status.Ready = false
 	case eks.NodegroupStatusUpdating:
 		managedPool.Status.Ready = true
+	case eks.NodegroupStatusDegraded:
+		issueErrMsgSet := make([]string, 0)
+		var errMsgStr string
+
+		for _, iss := range ng.Health.Issues {
+			errMsg := iss.GoString()
+			if slices.Contains(issueErrMsgSet, errMsg) {
+				continue
+			}
+			issueErrMsgSet = append(issueErrMsgSet, errMsg)
+			errMsgStr = fmt.Sprintf("%s %s", errMsgStr, errMsg)
+		}
+		reason := capierrors.InvalidConfigurationMachineError
+		// TODO: implement checks for other MachineStatusErrors and set reason accordingly
+		if strings.Contains(errMsgStr, "VcpuLimitExceeded") {
+			reason = capierrors.InsufficientResourcesMachineError
+		}
+
+		managedPool.Status.Ready = false
+		managedPool.Status.FailureReason = &reason
+		managedPool.Status.FailureMessage = &errMsgStr
+		return errors.Errorf("NodeGroup status is %s due to %v caused by error %s.  This error may persist and recreating the Node Group may be required to return to %s status",
+			eks.NodegroupStatusDegraded, *s.scope.ManagedMachinePool.Status.FailureReason, *s.scope.ManagedMachinePool.Status.FailureMessage, eks.NodegroupStatusActive)
 	default:
 		return errors.Errorf("unexpected EKS nodegroup status %s", *ng.Status)
 	}
diff --git a/pkg/cloud/services/eks/nodegroup_test.go b/pkg/cloud/services/eks/nodegroup_test.go
new file mode 100644
index 0000000000..89794fb99e
--- /dev/null
+++ b/pkg/cloud/services/eks/nodegroup_test.go
@@ -0,0 +1,48 @@
+package eks
+
+import (
+	"testing"
+
+	"github.com/aws/aws-sdk-go/service/eks"
+	. "github.com/onsi/gomega"
+	"sigs.k8s.io/cluster-api-provider-aws/v2/exp/api/v1beta2"
+	"sigs.k8s.io/cluster-api-provider-aws/v2/pkg/cloud/scope"
+	capierrors "sigs.k8s.io/cluster-api/errors"
+)
+
+func TestSetStatus(t *testing.T) {
+	g := NewWithT(t)
+	degraded := eks.NodegroupStatusDegraded
+	code := eks.NodegroupIssueCodeAsgInstanceLaunchFailures
+	message := "VcpuLimitExceeded"
+	resourceId := "my-worker-nodes"
+
+	s := &NodegroupService{
+		scope: &scope.ManagedMachinePoolScope{
+			ManagedMachinePool: &v1beta2.AWSManagedMachinePool{
+				Status: v1beta2.AWSManagedMachinePoolStatus{
+					Ready: false,
+				},
+			},
+		},
+	}
+
+	issue := &eks.Issue{
+		Code:        &code,
+		Message:     &message,
+		ResourceIds: []*string{&resourceId},
+	}
+	ng := &eks.Nodegroup{
+		Status: &degraded,
+		Health: &eks.NodegroupHealth{
+			Issues: []*eks.Issue{issue},
+		},
+	}
+
+	err := s.setStatus(ng)
+	g.Expect(err).ToNot(BeNil())
+	// ensure machine pool status values are set as expected
+	g.Expect(*s.scope.ManagedMachinePool.Status.FailureMessage).To(ContainSubstring(issue.GoString()))
+	g.Expect(s.scope.ManagedMachinePool.Status.Ready).To(Equal(false))
+	g.Expect(*s.scope.ManagedMachinePool.Status.FailureReason).To(Equal(capierrors.InsufficientResourcesMachineError))
+}