Skip to content

🌱 Add unit tests for failure domain in KCP #11687

New issue

Have a question about this project? Sign up for a free GitHub account to open an issue and contact its maintainers and the community.

Sign up for GitHub

By clicking “Sign up for GitHub”, you agree to our terms of service and privacy statement. We’ll occasionally send you account related emails.

Already on GitHub? Sign in to your account

Open
wants to merge 3 commits into
base: main
Choose a base branch
from
Open

🌱 Add unit tests for failure domain in KCP #11687

Show file tree
Hide file tree
Changes from all commits
Commits
Show all changes
3 commits
Select commit Hold shift + click to select a range
d373a7c
Add UTs for failure domain
arshadd-b Jan 15, 2025
753381d
address review comments
arshadd-b Jan 17, 2025
c58d175
fix failing uts
arshadd-b Apr 16, 2025
File filter

Filter by extension

Filter by extension
Conversations
Failed to load comments.
Loading
Jump to
Jump to file
Failed to load files.
Loading
Diff view
Diff view
263 changes: 263 additions & 0 deletions controlplane/kubeadm/internal/control_plane_test.go
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -24,13 +24,16 @@ import (
. "github.com/onsi/gomega"
corev1 "k8s.io/api/core/v1"
metav1 "k8s.io/apimachinery/pkg/apis/meta/v1"
"k8s.io/apimachinery/pkg/apis/meta/v1/unstructured"
"k8s.io/utils/ptr"
"sigs.k8s.io/controller-runtime/pkg/client"

clusterv1 "sigs.k8s.io/cluster-api/api/v1beta2"
controlplanev1 "sigs.k8s.io/cluster-api/controlplane/kubeadm/api/v1beta2"
"sigs.k8s.io/cluster-api/controlplane/kubeadm/internal/etcd"
"sigs.k8s.io/cluster-api/util/collections"
v1beta1conditions "sigs.k8s.io/cluster-api/util/conditions/deprecated/v1beta1"
"sigs.k8s.io/cluster-api/util/patch"
)

func TestControlPlane(t *testing.T) {
Expand Down Expand Up @@ -64,6 +67,12 @@ func TestControlPlane(t *testing.T) {
controlPlane.Machines.Insert(machine("machine-5", withFailureDomain("unknown")))
g.Expect(*controlPlane.FailureDomainWithMostMachines(ctx, controlPlane.Machines)).To(Equal("unknown"))
})

t.Run("With failure Domains is set empty", func(*testing.T) {
g := NewWithT(t)
controlPlane.Cluster.Status.FailureDomains = nil
g.Expect(*controlPlane.FailureDomainWithMostMachines(ctx, controlPlane.Machines)).To(Equal("one"))
})
})

t.Run("MachinesUpToDate", func(t *testing.T) {
Expand Down Expand Up @@ -144,6 +153,171 @@ func TestControlPlane(t *testing.T) {
g.Expect(err).NotTo(HaveOccurred())
g.Expect(fd).To(Equal(ptr.To("two"))) // deleted up-to-date machines (m4) should not be counted when picking the next failure domain for scale up
})

t.Run("Next Failure Domains", func(t *testing.T) {
Copy link
Contributor

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

This test is failing CI.

{Failed  === RUN   TestControlPlane/Next_Failure_Domains
    control_plane_test.go:181: 
        Unexpected error:
            <*errors.withStack | 0xc0009beae0>: 
            failed to retrieve infra obj for machine "m1": failed to retrieve GenericInfrastructureMachine m1: no matches for kind "GenericInfrastructureMachine" in version "infrastructure.cluster.x-k8s.io/v1beta1"
            {
                error: <*errors.withMessage | 0xc00131c4e0>{
                    cause: <*errors.withStack | 0xc0009beab0>{
                        error: <*errors.withMessage | 0xc00131c4c0>{
                            cause: <*meta.NoKindMatchError | 0xc00097ad00>{
                                GroupKind: {
                                    Group: "infrastructure.cluster.x-k8s.io",
                                    Kind: "GenericInfrastructureMachine",
                                },
                                SearchedVersions: ["v1beta1"],
                            },
                            msg: "failed to retrieve GenericInfrastructureMachine m1",
                        },
                        stack: [0x27453eb, 0x32d3805, 0x32d09bf, 0x338eeed, 0x61b5c7, 0x4b9621],
                    },
                    msg: "failed to retrieve infra obj for machine \"m1\"",
                },
                stack: [0x32d394d, 0x32d09bf, 0x338eeed, 0x61b5c7, 0x4b9621],
            }
        occurred
--- FAIL: TestControlPlane/Next_Failure_Domains (0.01s)
}

g := NewWithT(t)
cluster := clusterv1.Cluster{
Status: clusterv1.ClusterStatus{
FailureDomains: clusterv1.FailureDomains{
"one": failureDomain(false),
},
},
}
kcp := &controlplanev1.KubeadmControlPlane{
Spec: controlplanev1.KubeadmControlPlaneSpec{
Version: "v1.31.0",
},
}
machines := collections.Machines{
"machine-1": &clusterv1.Machine{
ObjectMeta: metav1.ObjectMeta{Name: "m1", DeletionTimestamp: ptr.To(metav1.Now())},
Spec: clusterv1.MachineSpec{
Version: ptr.To("v1.31.0"), // deleted
FailureDomain: ptr.To("one"),
InfrastructureRef: corev1.ObjectReference{Kind: "GenericInfrastructureMachine", APIVersion: clusterv1.GroupVersionInfrastructure.String(), Name: "m1"},
}},
}
controlPlane, err := NewControlPlane(ctx, nil, env.GetClient(), &cluster, kcp, machines)
g.Expect(err).NotTo(HaveOccurred())
fd, err := controlPlane.NextFailureDomainForScaleUp(ctx)
g.Expect(err).NotTo(HaveOccurred())
g.Expect(fd).To(BeNil())
})

t.Run("ControlPlane returns error when getting infra resources", func(t *testing.T) {
g := NewWithT(t)
cluster := clusterv1.Cluster{
Status: clusterv1.ClusterStatus{
FailureDomains: clusterv1.FailureDomains{
"one": failureDomain(true),
},
},
}
kcp := &controlplanev1.KubeadmControlPlane{
Spec: controlplanev1.KubeadmControlPlaneSpec{
Version: "v1.31.0",
},
}
machines := collections.Machines{
"machine-1": &clusterv1.Machine{
ObjectMeta: metav1.ObjectMeta{Name: "m1"},
Spec: clusterv1.MachineSpec{
Version: ptr.To("v1.31.0"),
FailureDomain: ptr.To("one"),
InfrastructureRef: corev1.ObjectReference{Name: "m1"},
}},
}
_, err := NewControlPlane(ctx, nil, env.GetClient(), &cluster, kcp, machines)
g.Expect(err).To(HaveOccurred())
})

t.Run("When infra and bootstrap config exists", func(t *testing.T) {
Copy link
Contributor

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

This case is failing in CI:

{Failed  === RUN   TestControlPlane/When_infra_and_bootstrap_config_exists
    control_plane_test.go:276: 
        Expected success, but got an error:
            <*meta.NoKindMatchError | 0xc001c9b140>: 
            no matches for kind "GenericInfrastructureMachine" in version "infrastructure.cluster.x-k8s.io/v1beta1"
            {
                GroupKind: {
                    Group: "infrastructure.cluster.x-k8s.io",
                    Kind: "GenericInfrastructureMachine",
                },
                SearchedVersions: ["v1beta1"],
            }
--- FAIL: TestControlPlane/When_infra_and_bootstrap_config_exists (0.01s)
}

Copy link
Contributor Author

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

will check, Thank you @nrb

g := NewWithT(t)
ns, err := env.CreateNamespace(ctx, "test-machine-watches")
kcp := &controlplanev1.KubeadmControlPlane{
Spec: controlplanev1.KubeadmControlPlaneSpec{
Version: "v1.31.0",
},
}

g.Expect(err).ToNot(HaveOccurred())

infraMachine := &unstructured.Unstructured{
Object: map[string]interface{}{
"kind": "GenericInfrastructureMachine",
"apiVersion": clusterv1.GroupVersionInfrastructure.String(),
"metadata": map[string]interface{}{
"name": "infra-config1",
"namespace": ns.Name,
},
"spec": map[string]interface{}{
"providerID": "test://id-1",
},
"status": map[string]interface{}{
"ready": true,
"addresses": []interface{}{
map[string]interface{}{
"type": "InternalIP",
"address": "10.0.0.1",
},
},
},
},
}

bootstrap := &unstructured.Unstructured{
Object: map[string]interface{}{
"kind": "KubeadmConfig",
"apiVersion": "bootstrap.cluster.x-k8s.io/v1beta1",
"metadata": map[string]interface{}{
"name": "bootstrap-config-machinereconcile",
"namespace": ns.Name,
},
"spec": map[string]interface{}{
"providerID": "test://id-1",
},
"status": map[string]interface{}{
"ready": true,
},
},
}

testCluster := &clusterv1.Cluster{
ObjectMeta: metav1.ObjectMeta{Name: "test-cluster", Namespace: ns.Name},
Status: clusterv1.ClusterStatus{
FailureDomains: clusterv1.FailureDomains{
"one": failureDomain(true),
"two": failureDomain(true),
"three": failureDomain(true),
},
},
}

g.Expect(env.Create(ctx, infraMachine)).To(Succeed())
g.Expect(env.Create(ctx, bootstrap)).To(Succeed())

defer func(do ...client.Object) {
g.Expect(env.Cleanup(ctx, do...)).To(Succeed())
}(ns, bootstrap, infraMachine)

// Patch infra machine ready
patchHelper, err := patch.NewHelper(infraMachine, env)
g.Expect(err).ShouldNot(HaveOccurred())
g.Expect(unstructured.SetNestedField(infraMachine.Object, true, "status", "ready")).To(Succeed())
g.Expect(patchHelper.Patch(ctx, infraMachine, patch.WithStatusObservedGeneration{})).To(Succeed())

// Patch bootstrap ready
patchHelper, err = patch.NewHelper(bootstrap, env)
g.Expect(err).ShouldNot(HaveOccurred())
g.Expect(unstructured.SetNestedField(bootstrap.Object, true, "status", "ready")).To(Succeed())
g.Expect(patchHelper.Patch(ctx, bootstrap, patch.WithStatusObservedGeneration{})).To(Succeed())

machines := collections.Machines{
"machine-1": &clusterv1.Machine{
ObjectMeta: metav1.ObjectMeta{Name: "m1",
Namespace: ns.Name},
Spec: clusterv1.MachineSpec{
InfrastructureRef: corev1.ObjectReference{
APIVersion: clusterv1.GroupVersionInfrastructure.String(),
Kind: "GenericInfrastructureMachine",
Name: "infra-config1",
Namespace: ns.Name,
},
Bootstrap: clusterv1.Bootstrap{
ConfigRef: &corev1.ObjectReference{
APIVersion: "bootstrap.cluster.x-k8s.io/v1beta1",
Kind: "KubeadmConfig",
Name: "bootstrap-config-machinereconcile",
Namespace: ns.Name,
},
},
},
},
}

_, err = NewControlPlane(ctx, nil, env.GetClient(), testCluster, kcp, machines)
g.Expect(err).NotTo(HaveOccurred())
})
}

func TestHasMachinesToBeRemediated(t *testing.T) {
Expand Down Expand Up @@ -332,6 +506,95 @@ func TestStatusToLogKeyAndValues(t *testing.T) {
g.Expect(got[3]).To(Equal("m1, m2, m3"))
}

func TestMachineInFailureDomainWithMostMachines(t *testing.T) {
t.Run("Machines in Failure Domain", func(t *testing.T) {
machines := collections.Machines{
"machine-3": &clusterv1.Machine{
ObjectMeta: metav1.ObjectMeta{Name: "m3"},
Spec: clusterv1.MachineSpec{
Version: ptr.To("v1.31.0"),
FailureDomain: ptr.To("three"),
InfrastructureRef: corev1.ObjectReference{Kind: "GenericInfrastructureMachine", APIVersion: "infrastructure.cluster.x-k8s.io/v1beta1", Name: "m3"},
}},
}

c := &ControlPlane{
KCP: &controlplanev1.KubeadmControlPlane{},
Cluster: &clusterv1.Cluster{
Status: clusterv1.ClusterStatus{
FailureDomains: clusterv1.FailureDomains{
"three": failureDomain(false),
},
},
},
Machines: collections.Machines{
"machine-3": machine("machine-3", withFailureDomain("three")),
},
}

g := NewWithT(t)
_, err := c.MachineInFailureDomainWithMostMachines(ctx, machines)
g.Expect(err).NotTo(HaveOccurred())
})
t.Run("Return error when no controlplane machine found", func(t *testing.T) {
machines := collections.Machines{}

c := &ControlPlane{
KCP: &controlplanev1.KubeadmControlPlane{},
Cluster: &clusterv1.Cluster{
Status: clusterv1.ClusterStatus{
FailureDomains: clusterv1.FailureDomains{},
},
},
Machines: collections.Machines{},
}

g := NewWithT(t)
_, err := c.MachineInFailureDomainWithMostMachines(ctx, machines)
g.Expect(err).To(HaveOccurred())
})
}
func TestMachineWithDeleteAnnotation(t *testing.T) {
t.Run("Machines having delete annotation set", func(t *testing.T) {
machines := collections.Machines{
"machine-1": &clusterv1.Machine{
ObjectMeta: metav1.ObjectMeta{Name: "m1",
Annotations: map[string]string{
"cluster.x-k8s.io/delete-machine": "",
},
},
Spec: clusterv1.MachineSpec{
Version: ptr.To("v1.31.0"),
FailureDomain: ptr.To("one"),
InfrastructureRef: corev1.ObjectReference{Kind: "GenericInfrastructureMachine", APIVersion: "infrastructure.cluster.x-k8s.io/v1beta1", Name: "m1"},
}},
"machine-2": &clusterv1.Machine{
ObjectMeta: metav1.ObjectMeta{Name: "m2",
Annotations: map[string]string{
"cluster.x-k8s.io/delete-machine": "",
},
},
Spec: clusterv1.MachineSpec{
Version: ptr.To("v1.31.0"),
FailureDomain: ptr.To("two"),
InfrastructureRef: corev1.ObjectReference{Kind: "GenericInfrastructureMachine", APIVersion: "infrastructure.cluster.x-k8s.io/v1beta1", Name: "m2"},
}},
}

c := ControlPlane{
Machines: machines,
Cluster: &clusterv1.Cluster{
Status: clusterv1.ClusterStatus{},
},
}

g := NewWithT(t)
annotatedMachines := c.MachineWithDeleteAnnotation(machines)
g.Expect(annotatedMachines).NotTo(BeNil())
g.Expect(annotatedMachines.Len()).To(BeEquivalentTo(2))
})
}

type machineOpt func(*clusterv1.Machine)

func failureDomain(controlPlane bool) clusterv1.FailureDomainSpec {
Expand Down
Loading