Skip to content

fix(scheduler): preserve concurrent node locks #2197

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

Merged
hami-robot merged 3 commits into from
Aug 3, 2026
+246 −7
Merged

fix(scheduler): preserve concurrent node locks #2197

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
2764a58
fix(scheduler): preserve concurrent node locks
ShiroKSH Jul 29, 2026
31aafbe
fix(scheduler): release restamped node locks
ShiroKSH Jul 30, 2026
e4c25be
fix(scheduler): preserve retry lock ownership
ShiroKSH Aug 1, 2026
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
33 changes: 28 additions & 5 deletions pkg/util/nodelock/nodelock.go
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -138,17 +138,24 @@ func SetNodeLock(nodeName string, lockname string, pods *corev1.Pod) error {
return err
}
if _, ok := node.Annotations[NodeLockKey]; ok {
return fmt.Errorf("node %s is locked", nodeName)
return fmt.Errorf("node %s is locked: %w", nodeName, ErrNodeLockContention)
}
lockOwner := NodeLockSep + GeneratePodNamespaceName(pods, NodeLockSep)
err = retry.OnError(DefaultStrategy, func(err error) bool {
// Retry on any error
return true
return !IsNodeLockContention(err)
}, func() error {
node, err = client.GetClient().CoreV1().Nodes().Get(ctx, nodeName, metav1.GetOptions{})
if err != nil {
klog.ErrorS(err, "Failed to get node when retry to patch", "node", nodeName)
return err
}
lockStr, ok := node.Annotations[NodeLockKey]
if ok && strings.Contains(lockStr, NodeLockSep) && strings.HasSuffix(lockStr, lockOwner) {
return nil
}
if ok {
return fmt.Errorf("node %s is locked: %w", nodeName, ErrNodeLockContention)
Comment thread
archlitchi marked this conversation as resolved.
}
patchData := fmt.Sprintf(`{"metadata":{"annotations":{"%s":"%s"},"resourceVersion":"%s"}}`, NodeLockKey, GenerateNodeLockKeyByPod(pods), node.ResourceVersion)
_, err = client.GetClient().CoreV1().Nodes().Patch(ctx, nodeName, types.MergePatchType, []byte(patchData), metav1.PatchOptions{})
if err != nil {
Expand Down Expand Up @@ -184,12 +191,14 @@ func ReleaseNodeLock(nodeName string, lockname string, pod *corev1.Pod, skipNode
if !ok {
return nil
}
lockOwner := NodeLockSep + GeneratePodNamespaceName(pod, NodeLockSep)
// Keep backward compatibility with the legacy format, which is simply a timestamp
if !skipNodeLockOwnerCheck && strings.Contains(lockStr, NodeLockSep) && !strings.HasSuffix(lockStr, NodeLockSep+GeneratePodNamespaceName(pod, NodeLockSep)) {
if !skipNodeLockOwnerCheck && strings.Contains(lockStr, NodeLockSep) && !strings.HasSuffix(lockStr, lockOwner) {
klog.InfoS("NodeLock is not set by this pod", NodeLockKey, lockStr, "podName", pod.Name, "podNamespace", pod.Namespace)
return nil
}

released := false
err = retry.OnError(DefaultStrategy, func(err error) bool {
// Retry on any error
return true
Expand All @@ -199,19 +208,33 @@ func ReleaseNodeLock(nodeName string, lockname string, pod *corev1.Pod, skipNode
klog.ErrorS(err, "Failed to get node when retry to patch", "node", nodeName)
return err
}
currentLock, ok := node.Annotations[NodeLockKey]
if !ok {
return nil
}
if skipNodeLockOwnerCheck || !strings.Contains(currentLock, NodeLockSep) {
if currentLock != lockStr {
return nil
}
} else if !strings.HasSuffix(currentLock, lockOwner) {
return nil
}
patchData := fmt.Sprintf(`{"metadata":{"annotations":{"%s":null},"resourceVersion":"%s"}}`, NodeLockKey, node.ResourceVersion)
_, err = client.GetClient().CoreV1().Nodes().Patch(ctx, nodeName, types.MergePatchType, []byte(patchData), metav1.PatchOptions{})
if err != nil {
klog.ErrorS(err, "Failed to patch node when retry to patch", "node", nodeName)
return err
}
released = true
return nil
})
if err != nil {
return fmt.Errorf("failed to release node lock (node=%s, retry strategy=%+v): %w", nodeName, DefaultStrategy, err)
}

klog.InfoS("Node lock released", "node", nodeName, "podName", pod.Name)
if released {
klog.InfoS("Node lock released", "node", nodeName, "podName", pod.Name)
}
return nil
}

Expand Down
220 changes: 218 additions & 2 deletions pkg/util/nodelock/nodelock_test.go
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -17,19 +17,235 @@ limitations under the License.
package nodelock

import (
"context" // Added for the new test
"context"
"errors"
"runtime"
"strings"
"testing"
"time"

corev1 "k8s.io/api/core/v1"
metav1 "k8s.io/apimachinery/pkg/apis/meta/v1" // Added for the new test
apierrors "k8s.io/apimachinery/pkg/api/errors"
metav1 "k8s.io/apimachinery/pkg/apis/meta/v1"
k8sruntime "k8s.io/apimachinery/pkg/runtime"
"k8s.io/apimachinery/pkg/runtime/schema"
"k8s.io/client-go/kubernetes/fake"
k8stesting "k8s.io/client-go/testing"

"github.com/Project-HAMi/HAMi/pkg/util/client"
)

func TestSetNodeLockPreservesConcurrentLockAfterConflict(t *testing.T) {
nodeLocks = newNodeLockManager()
nodeName := "node-set-conflict"
podA := &corev1.Pod{ObjectMeta: metav1.ObjectMeta{Name: "pod-a", Namespace: "ns"}}
podB := &corev1.Pod{ObjectMeta: metav1.ObjectMeta{Name: "pod-b", Namespace: "ns"}}
holderB := GenerateNodeLockKeyByPod(podB)
clientSet := fake.NewClientset(&corev1.Node{ObjectMeta: metav1.ObjectMeta{Name: nodeName}})
client.KubeClient = clientSet

getCalls := 0
clientSet.PrependReactor("get", "nodes", func(k8stesting.Action) (bool, k8sruntime.Object, error) {
getCalls++
if getCalls < 3 {
return false, nil, nil
}
return true, &corev1.Node{ObjectMeta: metav1.ObjectMeta{
Name: nodeName,
Annotations: map[string]string{NodeLockKey: holderB},
}}, nil
})
patchCalls := 0
clientSet.PrependReactor("patch", "nodes", func(k8stesting.Action) (bool, k8sruntime.Object, error) {
patchCalls++
return true, nil, apierrors.NewConflict(schema.GroupResource{Resource: "nodes"}, nodeName, errors.New("simulated concurrent lock"))
})

err := SetNodeLock(nodeName, "", podA)
if !IsNodeLockContention(err) {
t.Fatalf("SetNodeLock() error = %v, want node lock contention", err)
}
node, err := clientSet.CoreV1().Nodes().Get(context.Background(), nodeName, metav1.GetOptions{})
if err != nil {
t.Fatalf("get node: %v", err)
}
if got := node.Annotations[NodeLockKey]; got != holderB {
t.Fatalf("node lock = %q, want concurrent holder %q", got, holderB)
}
if patchCalls != 1 {
t.Fatalf("patch calls = %d, want 1", patchCalls)
}
}

func TestSetNodeLockSucceedsAfterLostPatchResponse(t *testing.T) {
nodeLocks = newNodeLockManager()
nodeName := "node-set-lost-response"
pod := &corev1.Pod{ObjectMeta: metav1.ObjectMeta{Name: "pod-a", Namespace: "ns"}}
lockStr := "2026-08-01T06:00:00Z,ns,pod-a"
clientSet := fake.NewClientset(&corev1.Node{ObjectMeta: metav1.ObjectMeta{Name: nodeName}})
client.KubeClient = clientSet

getCalls := 0
clientSet.PrependReactor("get", "nodes", func(k8stesting.Action) (bool, k8sruntime.Object, error) {
getCalls++
if getCalls < 3 {
return false, nil, nil
}
return true, &corev1.Node{ObjectMeta: metav1.ObjectMeta{
Name: nodeName,
Annotations: map[string]string{NodeLockKey: lockStr},
}}, nil
})
patchCalls := 0
clientSet.PrependReactor("patch", "nodes", func(k8stesting.Action) (bool, k8sruntime.Object, error) {
patchCalls++
return true, nil, errors.New("simulated lost patch response")
})

if err := SetNodeLock(nodeName, "", pod); err != nil {
t.Fatalf("SetNodeLock() error = %v, want nil", err)
}
if patchCalls != 1 {
t.Fatalf("patch calls = %d, want 1", patchCalls)
}
}

func TestReleaseNodeLockPreservesConcurrentLockAfterConflict(t *testing.T) {
nodeLocks = newNodeLockManager()
nodeName := "node-release-conflict"
podA := &corev1.Pod{ObjectMeta: metav1.ObjectMeta{Name: "pod-a", Namespace: "ns"}}
podB := &corev1.Pod{ObjectMeta: metav1.ObjectMeta{Name: "pod-b", Namespace: "ns"}}
holderA := GenerateNodeLockKeyByPod(podA)
holderB := GenerateNodeLockKeyByPod(podB)
clientSet := fake.NewClientset(&corev1.Node{ObjectMeta: metav1.ObjectMeta{
Name: nodeName,
Annotations: map[string]string{NodeLockKey: holderA},
}})
client.KubeClient = clientSet

getCalls := 0
clientSet.PrependReactor("get", "nodes", func(k8stesting.Action) (bool, k8sruntime.Object, error) {
getCalls++
if getCalls < 3 {
return false, nil, nil
}
return true, &corev1.Node{ObjectMeta: metav1.ObjectMeta{
Name: nodeName,
Annotations: map[string]string{NodeLockKey: holderB},
}}, nil
})
patchCalls := 0
clientSet.PrependReactor("patch", "nodes", func(k8stesting.Action) (bool, k8sruntime.Object, error) {
patchCalls++
return true, nil, apierrors.NewConflict(schema.GroupResource{Resource: "nodes"}, nodeName, errors.New("simulated concurrent lock"))
})

if err := ReleaseNodeLock(nodeName, "", podA, false); err != nil {
t.Fatalf("ReleaseNodeLock() error = %v, want nil", err)
}
node, err := clientSet.CoreV1().Nodes().Get(context.Background(), nodeName, metav1.GetOptions{})
if err != nil {
t.Fatalf("get node: %v", err)
}
if got := node.Annotations[NodeLockKey]; got != holderB {
t.Fatalf("node lock = %q, want concurrent holder %q", got, holderB)
}
if patchCalls != 1 {
t.Fatalf("patch calls = %d, want 1", patchCalls)
}
}

func TestReleaseNodeLockPreservesReplacedLegacyLockAfterConflict(t *testing.T) {
nodeLocks = newNodeLockManager()
nodeName := "node-release-legacy-conflict"
pod := &corev1.Pod{ObjectMeta: metav1.ObjectMeta{Name: "pod-a", Namespace: "ns"}}
initialLock := "2026-08-01T06:00:00Z"
replacedLock := "2026-08-01T06:00:01Z"
clientSet := fake.NewClientset(&corev1.Node{ObjectMeta: metav1.ObjectMeta{
Name: nodeName,
Annotations: map[string]string{NodeLockKey: initialLock},
}})
client.KubeClient = clientSet

getCalls := 0
clientSet.PrependReactor("get", "nodes", func(k8stesting.Action) (bool, k8sruntime.Object, error) {
getCalls++
if getCalls < 3 {
return false, nil, nil
}
return true, &corev1.Node{ObjectMeta: metav1.ObjectMeta{
Name: nodeName,
Annotations: map[string]string{NodeLockKey: replacedLock},
}}, nil
})
patchCalls := 0
clientSet.PrependReactor("patch", "nodes", func(k8stesting.Action) (bool, k8sruntime.Object, error) {
patchCalls++
return true, nil, apierrors.NewConflict(schema.GroupResource{Resource: "nodes"}, nodeName, errors.New("simulated concurrent legacy lock"))
})

if err := ReleaseNodeLock(nodeName, "", pod, false); err != nil {
t.Fatalf("ReleaseNodeLock() error = %v, want nil", err)
}
node, err := clientSet.CoreV1().Nodes().Get(context.Background(), nodeName, metav1.GetOptions{})
if err != nil {
t.Fatalf("get node: %v", err)
}
if got := node.Annotations[NodeLockKey]; got != replacedLock {
t.Fatalf("node lock = %q, want concurrent legacy lock %q", got, replacedLock)
}
if patchCalls != 1 {
t.Fatalf("patch calls = %d, want 1", patchCalls)
}
}

func TestReleaseNodeLockReleasesRestampedLockForSamePod(t *testing.T) {
nodeLocks = newNodeLockManager()
nodeName := "node-release-restamped"
pod := &corev1.Pod{ObjectMeta: metav1.ObjectMeta{Name: "pod-a", Namespace: "ns"}}
initialLock := "2026-07-29T13:00:00Z,ns,pod-a"
restampedLock := "2026-07-29T13:00:01Z,ns,pod-a"
clientSet := fake.NewClientset(&corev1.Node{ObjectMeta: metav1.ObjectMeta{
Name: nodeName,
Annotations: map[string]string{NodeLockKey: initialLock},
}})
client.KubeClient = clientSet

getCalls := 0
clientSet.PrependReactor("get", "nodes", func(k8stesting.Action) (bool, k8sruntime.Object, error) {
getCalls++
if getCalls != 3 {
return false, nil, nil
}
return true, &corev1.Node{ObjectMeta: metav1.ObjectMeta{
Name: nodeName,
Annotations: map[string]string{NodeLockKey: restampedLock},
}}, nil
})
patchCalls := 0
clientSet.PrependReactor("patch", "nodes", func(k8stesting.Action) (bool, k8sruntime.Object, error) {
patchCalls++
if patchCalls == 1 {
return true, nil, apierrors.NewConflict(schema.GroupResource{Resource: "nodes"}, nodeName, errors.New("simulated restamp"))
}
return false, nil, nil
})

if err := ReleaseNodeLock(nodeName, "", pod, false); err != nil {
t.Fatalf("ReleaseNodeLock() error = %v, want nil", err)
}
node, err := clientSet.CoreV1().Nodes().Get(context.Background(), nodeName, metav1.GetOptions{})
if err != nil {
t.Fatalf("get node: %v", err)
}
if _, ok := node.Annotations[NodeLockKey]; ok {
t.Fatalf("node lock = %q, want removed", node.Annotations[NodeLockKey])
}
if patchCalls != 2 {
t.Fatalf("patch calls = %d, want 2", patchCalls)
}
}

func Test_LockNode(t *testing.T) {
client.KubeClient = fake.NewClientset()
type args struct {
Expand Down
Loading