mirror of
https://github.com/hanzo-ml/kuberay.git
synced 2026-10-30 19:48:47 +00:00
* Fix dangling RayJob in HTTP mode * fix job status persistence * rename job status field * make ts record/check logic more general to all modes * fix inconsistent naming * comment on yaml * rename to check failure * re-generate * clear failure timestamp on success * refine job 404 * doc * simplify code * fix codegen
1275 lines
40 KiB
Go
1275 lines
40 KiB
Go
package ray
|
|
|
|
import (
|
|
"context"
|
|
"errors"
|
|
"fmt"
|
|
"math"
|
|
"os"
|
|
"strings"
|
|
"testing"
|
|
"time"
|
|
|
|
"github.com/stretchr/testify/assert"
|
|
"github.com/stretchr/testify/mock"
|
|
"github.com/stretchr/testify/require"
|
|
"go.uber.org/mock/gomock"
|
|
batchv1 "k8s.io/api/batch/v1"
|
|
corev1 "k8s.io/api/core/v1"
|
|
metav1 "k8s.io/apimachinery/pkg/apis/meta/v1"
|
|
"k8s.io/apimachinery/pkg/runtime"
|
|
"k8s.io/apimachinery/pkg/types"
|
|
"k8s.io/client-go/tools/record"
|
|
"k8s.io/utils/ptr"
|
|
"sigs.k8s.io/controller-runtime/pkg/client"
|
|
clientFake "sigs.k8s.io/controller-runtime/pkg/client/fake"
|
|
"sigs.k8s.io/controller-runtime/pkg/client/interceptor"
|
|
"sigs.k8s.io/controller-runtime/pkg/reconcile"
|
|
|
|
rayv1 "github.com/ray-project/kuberay/ray-operator/apis/ray/v1"
|
|
"github.com/ray-project/kuberay/ray-operator/controllers/ray/batchscheduler"
|
|
schedulerinterface "github.com/ray-project/kuberay/ray-operator/controllers/ray/batchscheduler/interface"
|
|
"github.com/ray-project/kuberay/ray-operator/controllers/ray/metrics/mocks"
|
|
utils "github.com/ray-project/kuberay/ray-operator/controllers/ray/utils"
|
|
"github.com/ray-project/kuberay/ray-operator/pkg/client/clientset/versioned/scheme"
|
|
"github.com/ray-project/kuberay/ray-operator/pkg/features"
|
|
)
|
|
|
|
// fakeBatchScheduler implements schedulerinterface.BatchScheduler for testing.
|
|
type fakeBatchScheduler struct {
|
|
schedulerinterface.DefaultBatchScheduler
|
|
cleanupCalled bool
|
|
cleanupObject metav1.Object
|
|
cleanupDidUpdate bool
|
|
cleanupErr error
|
|
}
|
|
|
|
func (f *fakeBatchScheduler) CleanupOnCompletion(_ context.Context, object metav1.Object) (bool, error) {
|
|
f.cleanupCalled = true
|
|
f.cleanupObject = object
|
|
return f.cleanupDidUpdate, f.cleanupErr
|
|
}
|
|
|
|
func TestCreateRayJobSubmitterIfNeed(t *testing.T) {
|
|
newScheme := runtime.NewScheme()
|
|
_ = rayv1.AddToScheme(newScheme)
|
|
_ = batchv1.AddToScheme(newScheme)
|
|
_ = corev1.AddToScheme(newScheme)
|
|
|
|
rayCluster := &rayv1.RayCluster{
|
|
ObjectMeta: metav1.ObjectMeta{
|
|
Name: "test-raycluster",
|
|
Namespace: "default",
|
|
},
|
|
Spec: rayv1.RayClusterSpec{
|
|
HeadGroupSpec: rayv1.HeadGroupSpec{
|
|
Template: corev1.PodTemplateSpec{
|
|
Spec: corev1.PodSpec{
|
|
Containers: []corev1.Container{
|
|
{
|
|
Image: "rayproject/ray",
|
|
},
|
|
},
|
|
},
|
|
},
|
|
},
|
|
},
|
|
}
|
|
|
|
rayJob := &rayv1.RayJob{
|
|
ObjectMeta: metav1.ObjectMeta{
|
|
Name: "test-rayjob",
|
|
Namespace: "default",
|
|
},
|
|
}
|
|
|
|
k8sJob := &batchv1.Job{
|
|
ObjectMeta: metav1.ObjectMeta{
|
|
Name: "test-rayjob",
|
|
Namespace: "default",
|
|
},
|
|
}
|
|
|
|
// Test 1: Return the existing k8s job if it already exists
|
|
fakeClient := clientFake.NewClientBuilder().WithScheme(newScheme).WithRuntimeObjects(k8sJob, rayCluster, rayJob).Build()
|
|
ctx := context.TODO()
|
|
|
|
rayJobReconciler := &RayJobReconciler{
|
|
Client: fakeClient,
|
|
Scheme: newScheme,
|
|
Recorder: &record.FakeRecorder{},
|
|
}
|
|
|
|
err := rayJobReconciler.createK8sJobIfNeed(ctx, rayJob, rayCluster)
|
|
require.NoError(t, err)
|
|
|
|
// Test 2: Create a new k8s job if it does not already exist
|
|
fakeClient = clientFake.NewClientBuilder().WithScheme(newScheme).WithRuntimeObjects(rayCluster, rayJob).Build()
|
|
rayJobReconciler.Client = fakeClient
|
|
|
|
err = rayJobReconciler.createK8sJobIfNeed(ctx, rayJob, rayCluster)
|
|
require.NoError(t, err)
|
|
|
|
err = fakeClient.Get(ctx, types.NamespacedName{
|
|
Namespace: k8sJob.Namespace,
|
|
Name: k8sJob.Name,
|
|
}, k8sJob, nil)
|
|
require.NoError(t, err)
|
|
|
|
assert.Equal(t, k8sJob.Labels[utils.RayOriginatedFromCRNameLabelKey], rayJob.Name)
|
|
assert.Equal(t, k8sJob.Labels[utils.RayOriginatedFromCRDLabelKey], utils.RayOriginatedFromCRDLabelValue(utils.RayJobCRD))
|
|
}
|
|
|
|
func TestGetSubmitterTemplate(t *testing.T) {
|
|
// RayJob instance with user-provided submitter pod template.
|
|
rayJobInstanceWithTemplate := &rayv1.RayJob{
|
|
Spec: rayv1.RayJobSpec{
|
|
Entrypoint: "echo no quote 'single quote' \"double quote\"",
|
|
SubmitterPodTemplate: &corev1.PodTemplateSpec{
|
|
Spec: corev1.PodSpec{
|
|
Containers: []corev1.Container{
|
|
{
|
|
Command: []string{"user-command"},
|
|
},
|
|
},
|
|
},
|
|
},
|
|
},
|
|
Status: rayv1.RayJobStatus{
|
|
DashboardURL: "test-url",
|
|
JobId: "test-job-id",
|
|
},
|
|
}
|
|
|
|
// RayJob instance without user-provided submitter pod template.
|
|
// In this case we should use the image of the Ray Head, so specify the image so we can test it.
|
|
rayJobInstanceWithoutTemplate := &rayv1.RayJob{
|
|
Spec: rayv1.RayJobSpec{
|
|
Entrypoint: "echo no quote 'single quote' \"double quote\"",
|
|
RayClusterSpec: &rayv1.RayClusterSpec{
|
|
HeadGroupSpec: rayv1.HeadGroupSpec{
|
|
Template: corev1.PodTemplateSpec{
|
|
Spec: corev1.PodSpec{
|
|
Containers: []corev1.Container{
|
|
{
|
|
Image: "rayproject/ray:custom-version",
|
|
},
|
|
},
|
|
},
|
|
},
|
|
},
|
|
},
|
|
},
|
|
Status: rayv1.RayJobStatus{
|
|
DashboardURL: "test-url",
|
|
JobId: "test-job-id",
|
|
},
|
|
}
|
|
rayClusterInstance := &rayv1.RayCluster{
|
|
Spec: rayv1.RayClusterSpec{
|
|
HeadGroupSpec: rayv1.HeadGroupSpec{
|
|
Template: corev1.PodTemplateSpec{
|
|
Spec: corev1.PodSpec{
|
|
Containers: []corev1.Container{
|
|
{
|
|
Image: "rayproject/ray:custom-version",
|
|
},
|
|
},
|
|
},
|
|
},
|
|
},
|
|
},
|
|
}
|
|
|
|
// Test 1: User provided template with command
|
|
submitterTemplate, err := getSubmitterTemplate(rayJobInstanceWithTemplate, rayClusterInstance)
|
|
require.NoError(t, err)
|
|
assert.Equal(t, "user-command", submitterTemplate.Spec.Containers[utils.RayContainerIndex].Command[0])
|
|
|
|
// Test 2: User provided template without command
|
|
rayJobInstanceWithTemplate.Spec.SubmitterPodTemplate.Spec.Containers[utils.RayContainerIndex].Command = []string{}
|
|
submitterTemplate, err = getSubmitterTemplate(rayJobInstanceWithTemplate, rayClusterInstance)
|
|
require.NoError(t, err)
|
|
assert.Equal(t, []string{"/bin/bash", "-ce", "--"}, submitterTemplate.Spec.Containers[utils.RayContainerIndex].Command)
|
|
expectedK8sJobModeArgs := []string{
|
|
"until " + fmt.Sprintf(utils.BasePythonHealthCommand, "http://test-url/"+utils.RayDashboardGCSHealthPath, utils.RayDashboardGCSHealthCheckTimeoutSeconds) +
|
|
" >/dev/null 2>&1 ; do echo \"Waiting for Ray Dashboard GCS to become healthy at http://test-url ...\" ; sleep 2 ; done ; " +
|
|
"if ! ray job status --address http://test-url test-job-id >/dev/null 2>&1 ; then ray job submit --address http://test-url --no-wait --submission-id test-job-id -- echo no quote 'single quote' \"double quote\" ; fi ; ray job logs --address http://test-url --follow test-job-id",
|
|
}
|
|
assert.Equal(t, expectedK8sJobModeArgs, submitterTemplate.Spec.Containers[utils.RayContainerIndex].Args)
|
|
|
|
// Test 3: User did not provide template, should use the image of the Ray Head
|
|
submitterTemplate, err = getSubmitterTemplate(rayJobInstanceWithoutTemplate, rayClusterInstance)
|
|
require.NoError(t, err)
|
|
assert.Equal(t, []string{"/bin/bash", "-ce", "--"}, submitterTemplate.Spec.Containers[utils.RayContainerIndex].Command)
|
|
assert.Equal(t, expectedK8sJobModeArgs, submitterTemplate.Spec.Containers[utils.RayContainerIndex].Args)
|
|
assert.Equal(t, "rayproject/ray:custom-version", submitterTemplate.Spec.Containers[utils.RayContainerIndex].Image)
|
|
|
|
// Test 4: Check default PYTHONUNBUFFERED setting
|
|
submitterTemplate, err = getSubmitterTemplate(rayJobInstanceWithoutTemplate, rayClusterInstance)
|
|
require.NoError(t, err)
|
|
|
|
envVar, found := utils.EnvVarByName(PythonUnbufferedEnvVarName, submitterTemplate.Spec.Containers[utils.RayContainerIndex].Env)
|
|
assert.True(t, found)
|
|
assert.Equal(t, "1", envVar.Value)
|
|
|
|
// Test 5: Check default RAY_DASHBOARD_ADDRESS env var
|
|
submitterTemplate, err = getSubmitterTemplate(rayJobInstanceWithTemplate, rayClusterInstance)
|
|
require.NoError(t, err)
|
|
|
|
envVar, found = utils.EnvVarByName(utils.RAY_DASHBOARD_ADDRESS, submitterTemplate.Spec.Containers[utils.RayContainerIndex].Env)
|
|
assert.True(t, found)
|
|
assert.Equal(t, "test-url", envVar.Value)
|
|
|
|
// Test 6: Check default RAY_JOB_SUBMISSION_ID env var
|
|
envVar, found = utils.EnvVarByName(utils.RAY_JOB_SUBMISSION_ID, submitterTemplate.Spec.Containers[utils.RayContainerIndex].Env)
|
|
assert.True(t, found)
|
|
assert.Equal(t, "test-job-id", envVar.Value)
|
|
}
|
|
|
|
func TestGetSubmitterContainerWithFeatureGate(t *testing.T) {
|
|
// Enable the SidecarSubmitterRestart feature gate for this test
|
|
features.SetFeatureGateDuringTest(t, features.SidecarSubmitterRestart, true)
|
|
|
|
rayJobInstance := &rayv1.RayJob{
|
|
Spec: rayv1.RayJobSpec{
|
|
Entrypoint: "echo test",
|
|
SubmissionMode: rayv1.SidecarMode,
|
|
RayClusterSpec: &rayv1.RayClusterSpec{
|
|
HeadGroupSpec: rayv1.HeadGroupSpec{
|
|
Template: corev1.PodTemplateSpec{
|
|
Spec: corev1.PodSpec{
|
|
Containers: []corev1.Container{
|
|
{
|
|
Image: "rayproject/ray:test",
|
|
Ports: []corev1.ContainerPort{
|
|
{
|
|
Name: utils.DashboardPortName,
|
|
ContainerPort: utils.DefaultDashboardPort,
|
|
},
|
|
},
|
|
},
|
|
},
|
|
},
|
|
},
|
|
},
|
|
},
|
|
},
|
|
Status: rayv1.RayJobStatus{
|
|
DashboardURL: "http://127.0.0.1:8265",
|
|
JobId: "test-job-id",
|
|
},
|
|
}
|
|
|
|
rayClusterInstance := &rayv1.RayCluster{
|
|
Spec: *rayJobInstance.Spec.RayClusterSpec,
|
|
}
|
|
|
|
container, err := getSubmitterContainer(rayJobInstance, rayClusterInstance)
|
|
require.NoError(t, err)
|
|
|
|
// Verify restart policy is set to OnFailure for the submitter container
|
|
require.NotNil(t, container.RestartPolicy)
|
|
assert.Equal(t, corev1.ContainerRestartPolicyOnFailure, *container.RestartPolicy)
|
|
}
|
|
|
|
func TestCheckIsRestartCountExceeded(t *testing.T) {
|
|
tests := []struct {
|
|
name string
|
|
restartCount int32
|
|
existCode int32
|
|
hasTermState bool
|
|
hasLastTermState bool
|
|
expectedShouldUpdate bool
|
|
}{
|
|
{
|
|
name: "restart count below limit of 2",
|
|
restartCount: 1,
|
|
existCode: 1,
|
|
hasTermState: true,
|
|
hasLastTermState: true,
|
|
expectedShouldUpdate: false,
|
|
},
|
|
{
|
|
name: "restart count at limit of 2",
|
|
restartCount: 2,
|
|
existCode: 1,
|
|
hasTermState: true,
|
|
hasLastTermState: true,
|
|
expectedShouldUpdate: true,
|
|
},
|
|
{
|
|
name: "successful exit should not trigger failure",
|
|
restartCount: 3,
|
|
existCode: 0,
|
|
hasTermState: true,
|
|
hasLastTermState: true,
|
|
expectedShouldUpdate: false,
|
|
},
|
|
}
|
|
|
|
for _, tt := range tests {
|
|
t.Run(tt.name, func(t *testing.T) {
|
|
containerStatus := corev1.ContainerStatus{
|
|
Name: utils.SubmitterContainerName,
|
|
RestartCount: tt.restartCount,
|
|
}
|
|
|
|
if tt.hasTermState {
|
|
containerStatus.State = corev1.ContainerState{
|
|
Terminated: &corev1.ContainerStateTerminated{ExitCode: tt.existCode},
|
|
}
|
|
}
|
|
|
|
if tt.hasLastTermState {
|
|
containerStatus.LastTerminationState = corev1.ContainerState{
|
|
Terminated: &corev1.ContainerStateTerminated{ExitCode: 1},
|
|
}
|
|
}
|
|
|
|
headPod := &corev1.Pod{
|
|
Status: corev1.PodStatus{
|
|
ContainerStatuses: []corev1.ContainerStatus{containerStatus},
|
|
},
|
|
}
|
|
|
|
shouldUpdate, _ := checkIsRestartCountExceeded(headPod, 2)
|
|
assert.Equal(t, tt.expectedShouldUpdate, shouldUpdate)
|
|
})
|
|
}
|
|
}
|
|
|
|
func TestUpdateStatusToSuspendingIfNeeded(t *testing.T) {
|
|
newScheme := runtime.NewScheme()
|
|
_ = rayv1.AddToScheme(newScheme)
|
|
tests := []struct {
|
|
name string
|
|
status rayv1.JobDeploymentStatus
|
|
suspend bool
|
|
expectedShouldUpdate bool
|
|
}{
|
|
// When Autoscaler is enabled, the random Pod deletion is controleld by the feature flag `ENABLE_RANDOM_POD_DELETE`.
|
|
{
|
|
name: "Suspend is false",
|
|
suspend: false,
|
|
status: rayv1.JobDeploymentStatusInitializing,
|
|
expectedShouldUpdate: false,
|
|
},
|
|
{
|
|
name: "Suspend is true, but the status is not allowed to transition to suspending",
|
|
suspend: true,
|
|
status: rayv1.JobDeploymentStatusComplete,
|
|
expectedShouldUpdate: false,
|
|
},
|
|
{
|
|
name: "Suspend is true, and the status is allowed to transition to suspending",
|
|
suspend: true,
|
|
status: rayv1.JobDeploymentStatusInitializing,
|
|
expectedShouldUpdate: true,
|
|
},
|
|
}
|
|
|
|
for _, tc := range tests {
|
|
t.Run(tc.name, func(t *testing.T) {
|
|
name := "test-rayjob"
|
|
namespace := "default"
|
|
rayJob := &rayv1.RayJob{
|
|
ObjectMeta: metav1.ObjectMeta{
|
|
Name: name,
|
|
Namespace: namespace,
|
|
},
|
|
Spec: rayv1.RayJobSpec{
|
|
Suspend: tc.suspend,
|
|
},
|
|
Status: rayv1.RayJobStatus{
|
|
JobDeploymentStatus: tc.status,
|
|
},
|
|
}
|
|
|
|
ctx := context.Background()
|
|
shouldUpdate := updateStatusToSuspendingIfNeeded(ctx, rayJob)
|
|
assert.Equal(t, tc.expectedShouldUpdate, shouldUpdate)
|
|
|
|
if tc.expectedShouldUpdate {
|
|
assert.Equal(t, rayv1.JobDeploymentStatusSuspending, rayJob.Status.JobDeploymentStatus)
|
|
} else {
|
|
assert.Equal(t, tc.status, rayJob.Status.JobDeploymentStatus)
|
|
}
|
|
})
|
|
}
|
|
}
|
|
|
|
func TestUpdateRayJobStatus(t *testing.T) {
|
|
newScheme := runtime.NewScheme()
|
|
_ = rayv1.AddToScheme(newScheme)
|
|
|
|
rayJobTemplate := &rayv1.RayJob{
|
|
ObjectMeta: metav1.ObjectMeta{
|
|
Name: "test-rayjob",
|
|
Namespace: "default",
|
|
},
|
|
Status: rayv1.RayJobStatus{
|
|
JobDeploymentStatus: rayv1.JobDeploymentStatusRunning,
|
|
JobStatus: rayv1.JobStatusRunning,
|
|
Message: "old message",
|
|
},
|
|
}
|
|
newMessage := "new message"
|
|
|
|
tests := []struct {
|
|
name string
|
|
isJobDeploymentStatusChanged bool
|
|
}{
|
|
{
|
|
name: "JobDeploymentStatus is not changed",
|
|
isJobDeploymentStatusChanged: false,
|
|
},
|
|
{
|
|
name: "JobDeploymentStatus is changed",
|
|
isJobDeploymentStatusChanged: true,
|
|
},
|
|
}
|
|
|
|
for _, tc := range tests {
|
|
t.Run(tc.name, func(t *testing.T) {
|
|
oldRayJob := rayJobTemplate.DeepCopy()
|
|
|
|
// Initialize a fake client with newScheme and runtimeObjects.
|
|
fakeClient := clientFake.NewClientBuilder().
|
|
WithScheme(newScheme).
|
|
WithRuntimeObjects(oldRayJob).
|
|
WithStatusSubresource(oldRayJob).Build()
|
|
ctx := context.Background()
|
|
|
|
newRayJob := &rayv1.RayJob{}
|
|
err := fakeClient.Get(ctx, types.NamespacedName{Namespace: oldRayJob.Namespace, Name: oldRayJob.Name}, newRayJob)
|
|
require.NoError(t, err)
|
|
|
|
// Update the status
|
|
newRayJob.Status.Message = newMessage
|
|
if tc.isJobDeploymentStatusChanged {
|
|
newRayJob.Status.JobDeploymentStatus = rayv1.JobDeploymentStatusSuspending
|
|
}
|
|
|
|
// Initialize a new RayClusterReconciler.
|
|
testRayJobReconciler := &RayJobReconciler{
|
|
Client: fakeClient,
|
|
Recorder: &record.FakeRecorder{},
|
|
Scheme: newScheme,
|
|
}
|
|
|
|
err = testRayJobReconciler.updateRayJobStatus(ctx, oldRayJob, newRayJob)
|
|
require.NoError(t, err)
|
|
|
|
err = fakeClient.Get(ctx, types.NamespacedName{Namespace: newRayJob.Namespace, Name: newRayJob.Name}, newRayJob)
|
|
require.NoError(t, err)
|
|
assert.Equal(t, newRayJob.Status.Message == newMessage, tc.isJobDeploymentStatusChanged)
|
|
})
|
|
}
|
|
}
|
|
|
|
func TestUpdateRayJobStatusPersistsJobStatusCheckFailureStartTime(t *testing.T) {
|
|
newScheme := runtime.NewScheme()
|
|
require.NoError(t, rayv1.AddToScheme(newScheme))
|
|
|
|
oldRayJob := &rayv1.RayJob{
|
|
ObjectMeta: metav1.ObjectMeta{
|
|
Name: "test-rayjob",
|
|
Namespace: "default",
|
|
},
|
|
Status: rayv1.RayJobStatus{
|
|
JobDeploymentStatus: rayv1.JobDeploymentStatusRunning,
|
|
JobStatus: rayv1.JobStatusRunning,
|
|
},
|
|
}
|
|
|
|
fakeClient := clientFake.NewClientBuilder().
|
|
WithScheme(newScheme).
|
|
WithRuntimeObjects(oldRayJob).
|
|
WithStatusSubresource(oldRayJob).Build()
|
|
ctx := context.Background()
|
|
|
|
newRayJob := &rayv1.RayJob{}
|
|
err := fakeClient.Get(ctx, types.NamespacedName{Namespace: oldRayJob.Namespace, Name: oldRayJob.Name}, newRayJob)
|
|
require.NoError(t, err)
|
|
|
|
startTime := metav1.NewTime(time.Now())
|
|
newRayJob.Status.JobStatusCheckFailureStartTime = &startTime
|
|
|
|
testRayJobReconciler := &RayJobReconciler{
|
|
Client: fakeClient,
|
|
Recorder: &record.FakeRecorder{},
|
|
Scheme: newScheme,
|
|
}
|
|
|
|
err = testRayJobReconciler.updateRayJobStatus(ctx, oldRayJob, newRayJob)
|
|
require.NoError(t, err)
|
|
|
|
err = fakeClient.Get(ctx, types.NamespacedName{Namespace: newRayJob.Namespace, Name: newRayJob.Name}, newRayJob)
|
|
require.NoError(t, err)
|
|
require.NotNil(t, newRayJob.Status.JobStatusCheckFailureStartTime)
|
|
assert.WithinDuration(t, startTime.Time, newRayJob.Status.JobStatusCheckFailureStartTime.Time, time.Second)
|
|
}
|
|
|
|
func TestUpdateRayJobStatusClearsJobStatusCheckFailureStartTime(t *testing.T) {
|
|
newScheme := runtime.NewScheme()
|
|
require.NoError(t, rayv1.AddToScheme(newScheme))
|
|
|
|
startTime := metav1.NewTime(time.Now())
|
|
oldRayJob := &rayv1.RayJob{
|
|
ObjectMeta: metav1.ObjectMeta{
|
|
Name: "test-rayjob",
|
|
Namespace: "default",
|
|
},
|
|
Status: rayv1.RayJobStatus{
|
|
JobDeploymentStatus: rayv1.JobDeploymentStatusRunning,
|
|
JobStatus: rayv1.JobStatusRunning,
|
|
JobStatusCheckFailureStartTime: &startTime,
|
|
},
|
|
}
|
|
|
|
fakeClient := clientFake.NewClientBuilder().
|
|
WithScheme(newScheme).
|
|
WithRuntimeObjects(oldRayJob).
|
|
WithStatusSubresource(oldRayJob).Build()
|
|
ctx := context.Background()
|
|
|
|
newRayJob := &rayv1.RayJob{}
|
|
err := fakeClient.Get(ctx, types.NamespacedName{Namespace: oldRayJob.Namespace, Name: oldRayJob.Name}, newRayJob)
|
|
require.NoError(t, err)
|
|
newRayJob.Status.JobStatusCheckFailureStartTime = nil
|
|
|
|
testRayJobReconciler := &RayJobReconciler{
|
|
Client: fakeClient,
|
|
Recorder: &record.FakeRecorder{},
|
|
Scheme: newScheme,
|
|
}
|
|
|
|
err = testRayJobReconciler.updateRayJobStatus(ctx, oldRayJob, newRayJob)
|
|
require.NoError(t, err)
|
|
|
|
err = fakeClient.Get(ctx, types.NamespacedName{Namespace: newRayJob.Namespace, Name: newRayJob.Name}, newRayJob)
|
|
require.NoError(t, err)
|
|
assert.Nil(t, newRayJob.Status.JobStatusCheckFailureStartTime)
|
|
}
|
|
|
|
func TestFailedToCreateRayJobSubmitterEvent(t *testing.T) {
|
|
rayJob := &rayv1.RayJob{
|
|
ObjectMeta: metav1.ObjectMeta{
|
|
Name: "test-rayjob",
|
|
Namespace: "default",
|
|
},
|
|
}
|
|
|
|
submitterTemplate := corev1.PodTemplateSpec{
|
|
ObjectMeta: metav1.ObjectMeta{
|
|
Name: "test-submit-pod",
|
|
Namespace: "default",
|
|
},
|
|
Spec: corev1.PodSpec{
|
|
Containers: []corev1.Container{
|
|
{
|
|
Name: "ray-submit",
|
|
Image: "rayproject/ray:latest",
|
|
},
|
|
},
|
|
},
|
|
}
|
|
|
|
fakeClient := clientFake.NewClientBuilder().WithInterceptorFuncs(interceptor.Funcs{
|
|
Create: func(_ context.Context, _ client.WithWatch, _ client.Object, _ ...client.CreateOption) error {
|
|
return errors.New("random")
|
|
},
|
|
}).WithScheme(scheme.Scheme).Build()
|
|
|
|
recorder := record.NewFakeRecorder(100)
|
|
|
|
reconciler := &RayJobReconciler{
|
|
Client: fakeClient,
|
|
Recorder: recorder,
|
|
Scheme: scheme.Scheme,
|
|
}
|
|
|
|
err := reconciler.createNewK8sJob(context.Background(), rayJob, submitterTemplate)
|
|
|
|
require.Error(t, err, "Expected error due to simulated job creation failure")
|
|
|
|
var foundFailureEvent bool
|
|
events := []string{}
|
|
for len(recorder.Events) > 0 {
|
|
event := <-recorder.Events
|
|
if strings.Contains(event, "Failed to create new Kubernetes Job") {
|
|
foundFailureEvent = true
|
|
break
|
|
}
|
|
events = append(events, event)
|
|
}
|
|
|
|
assert.Truef(t, foundFailureEvent, "Expected event to be generated for job creation failure, got events: %s", strings.Join(events, "\n"))
|
|
}
|
|
|
|
// TestCreateNewK8sJob_PropagatesLabelsToSubmitterPodTemplate verifies that createNewK8sJob
|
|
// propagates KubeRay labels to the submitter pod template so that NetworkPolicies can
|
|
// select submitter pods by KubeRay-specific labels.
|
|
func TestCreateNewK8sJob_PropagatesLabelsToSubmitterPodTemplate(t *testing.T) {
|
|
rayJob := &rayv1.RayJob{
|
|
ObjectMeta: metav1.ObjectMeta{
|
|
Name: "test-rayjob",
|
|
Namespace: "default",
|
|
UID: "test-uid",
|
|
},
|
|
}
|
|
|
|
submitterTemplate := corev1.PodTemplateSpec{
|
|
ObjectMeta: metav1.ObjectMeta{
|
|
Labels: map[string]string{
|
|
utils.RayOriginatedFromCRNameLabelKey: rayJob.Name,
|
|
utils.RayOriginatedFromCRDLabelKey: utils.RayOriginatedFromCRDLabelValue(utils.RayJobCRD),
|
|
utils.KubernetesCreatedByLabelKey: utils.ComponentName,
|
|
},
|
|
},
|
|
Spec: corev1.PodSpec{
|
|
Containers: []corev1.Container{
|
|
{Name: "ray-submit", Image: "rayproject/ray:latest"},
|
|
},
|
|
},
|
|
}
|
|
|
|
var createdJob *batchv1.Job
|
|
fakeClient := clientFake.NewClientBuilder().WithInterceptorFuncs(interceptor.Funcs{
|
|
Create: func(_ context.Context, _ client.WithWatch, obj client.Object, _ ...client.CreateOption) error {
|
|
if job, ok := obj.(*batchv1.Job); ok {
|
|
createdJob = job
|
|
}
|
|
return nil
|
|
},
|
|
}).WithScheme(scheme.Scheme).Build()
|
|
|
|
reconciler := &RayJobReconciler{
|
|
Client: fakeClient,
|
|
Recorder: record.NewFakeRecorder(10),
|
|
Scheme: scheme.Scheme,
|
|
}
|
|
|
|
err := reconciler.createNewK8sJob(context.Background(), rayJob, submitterTemplate)
|
|
require.NoError(t, err)
|
|
require.NotNil(t, createdJob)
|
|
|
|
podLabels := createdJob.Spec.Template.Labels
|
|
assert.Equal(t, "test-rayjob", podLabels[utils.RayOriginatedFromCRNameLabelKey],
|
|
"Pod template should have ray.io/originated-from-cr-name label")
|
|
assert.Equal(t, utils.RayOriginatedFromCRDLabelValue(utils.RayJobCRD), podLabels[utils.RayOriginatedFromCRDLabelKey],
|
|
"Pod template should have ray.io/originated-from-crd label")
|
|
assert.Equal(t, utils.ComponentName, podLabels[utils.KubernetesCreatedByLabelKey],
|
|
"Pod template should have app.kubernetes.io/created-by label")
|
|
}
|
|
|
|
func TestFailedCreateRayClusterEvent(t *testing.T) {
|
|
rayJob := &rayv1.RayJob{
|
|
ObjectMeta: metav1.ObjectMeta{
|
|
Name: "test-rayjob",
|
|
Namespace: "default",
|
|
},
|
|
Spec: rayv1.RayJobSpec{
|
|
RayClusterSpec: &rayv1.RayClusterSpec{},
|
|
},
|
|
}
|
|
|
|
fakeClient := clientFake.NewClientBuilder().WithInterceptorFuncs(interceptor.Funcs{
|
|
Create: func(_ context.Context, _ client.WithWatch, _ client.Object, _ ...client.CreateOption) error {
|
|
return errors.New("random")
|
|
},
|
|
}).WithScheme(scheme.Scheme).Build()
|
|
|
|
recorder := record.NewFakeRecorder(100)
|
|
|
|
reconciler := &RayJobReconciler{
|
|
Client: fakeClient,
|
|
Recorder: recorder,
|
|
Scheme: scheme.Scheme,
|
|
}
|
|
|
|
_, err := reconciler.getOrCreateRayClusterInstance(context.Background(), rayJob)
|
|
|
|
require.Error(t, err, "Expected error due to cluster creation failure")
|
|
|
|
var foundFailureEvent bool
|
|
events := []string{}
|
|
for len(recorder.Events) > 0 {
|
|
event := <-recorder.Events
|
|
if strings.Contains(event, "Failed to create RayCluster") {
|
|
foundFailureEvent = true
|
|
break
|
|
}
|
|
events = append(events, event)
|
|
}
|
|
|
|
assert.Truef(t, foundFailureEvent, "Expected event to be generated for cluster creation failure, got events: %s", strings.Join(events, "\n"))
|
|
}
|
|
|
|
func TestFailedDeleteRayJobSubmitterEvent(t *testing.T) {
|
|
newScheme := runtime.NewScheme()
|
|
_ = batchv1.AddToScheme(newScheme)
|
|
|
|
rayJob := &rayv1.RayJob{
|
|
ObjectMeta: metav1.ObjectMeta{
|
|
Name: "test-rayjob",
|
|
Namespace: "default",
|
|
},
|
|
}
|
|
submitter := &batchv1.Job{
|
|
ObjectMeta: metav1.ObjectMeta{
|
|
Name: "test-rayjob",
|
|
Namespace: "default",
|
|
},
|
|
}
|
|
|
|
fakeClient := clientFake.NewClientBuilder().WithInterceptorFuncs(interceptor.Funcs{
|
|
Delete: func(_ context.Context, _ client.WithWatch, _ client.Object, _ ...client.DeleteOption) error {
|
|
return errors.New("random")
|
|
},
|
|
}).WithScheme(newScheme).WithRuntimeObjects(submitter).Build()
|
|
|
|
recorder := record.NewFakeRecorder(100)
|
|
|
|
reconciler := &RayJobReconciler{
|
|
Client: fakeClient,
|
|
Recorder: recorder,
|
|
Scheme: scheme.Scheme,
|
|
}
|
|
|
|
_, err := reconciler.deleteSubmitterJob(context.Background(), rayJob)
|
|
|
|
require.Error(t, err, "Expected error due to job deletion failure")
|
|
|
|
var foundFailureEvent bool
|
|
events := []string{}
|
|
for len(recorder.Events) > 0 {
|
|
event := <-recorder.Events
|
|
if strings.Contains(event, "Failed to delete submitter K8s Job") {
|
|
foundFailureEvent = true
|
|
break
|
|
}
|
|
events = append(events, event)
|
|
}
|
|
|
|
assert.Truef(t, foundFailureEvent, "Expected event to be generated for cluster deletion failure, got events: %s", strings.Join(events, "\n"))
|
|
}
|
|
|
|
func TestFailedDeleteRayClusterEvent(t *testing.T) {
|
|
newScheme := runtime.NewScheme()
|
|
_ = rayv1.AddToScheme(newScheme)
|
|
|
|
rayCluster := &rayv1.RayCluster{
|
|
ObjectMeta: metav1.ObjectMeta{
|
|
Name: "test-raycluster",
|
|
Namespace: "default",
|
|
},
|
|
}
|
|
|
|
rayJob := &rayv1.RayJob{
|
|
ObjectMeta: metav1.ObjectMeta{
|
|
Name: "test-rayjob",
|
|
Namespace: "default",
|
|
},
|
|
Status: rayv1.RayJobStatus{
|
|
RayClusterName: "test-raycluster",
|
|
},
|
|
}
|
|
|
|
fakeClient := clientFake.NewClientBuilder().WithInterceptorFuncs(interceptor.Funcs{
|
|
Delete: func(_ context.Context, _ client.WithWatch, _ client.Object, _ ...client.DeleteOption) error {
|
|
return errors.New("random")
|
|
},
|
|
}).WithScheme(newScheme).WithRuntimeObjects(rayCluster).Build()
|
|
|
|
recorder := record.NewFakeRecorder(100)
|
|
|
|
reconciler := &RayJobReconciler{
|
|
Client: fakeClient,
|
|
Recorder: recorder,
|
|
Scheme: scheme.Scheme,
|
|
}
|
|
|
|
_, err := reconciler.deleteClusterResources(context.Background(), rayJob)
|
|
|
|
require.Error(t, err, "Expected error due to cluster deletion failure")
|
|
|
|
var foundFailureEvent bool
|
|
events := []string{}
|
|
for len(recorder.Events) > 0 {
|
|
event := <-recorder.Events
|
|
if strings.Contains(event, "Failed to delete cluster") {
|
|
foundFailureEvent = true
|
|
break
|
|
}
|
|
events = append(events, event)
|
|
}
|
|
|
|
assert.Truef(t, foundFailureEvent, "Expected event to be generated for cluster deletion failure, got events: %s", strings.Join(events, "\n"))
|
|
}
|
|
|
|
func TestEmitRayJobExecutionDuration(t *testing.T) {
|
|
rayJobName := "test-job"
|
|
rayJobNamespace := "default"
|
|
rayJobUID := types.UID("test-job-uid")
|
|
mockTime := time.Now().Add(-60 * time.Second)
|
|
|
|
tests := []struct {
|
|
name string
|
|
originalRayJobStatus rayv1.RayJobStatus
|
|
rayJobStatus rayv1.RayJobStatus
|
|
expectMetricsCall bool
|
|
expectedJobDeploymentStatus rayv1.JobDeploymentStatus
|
|
expectedRetryCount int
|
|
expectedDuration float64
|
|
}{
|
|
{
|
|
name: "non-terminal to complete state should emit metrics",
|
|
originalRayJobStatus: rayv1.RayJobStatus{
|
|
JobDeploymentStatus: rayv1.JobDeploymentStatusRunning,
|
|
},
|
|
rayJobStatus: rayv1.RayJobStatus{
|
|
JobDeploymentStatus: rayv1.JobDeploymentStatusComplete,
|
|
StartTime: &metav1.Time{Time: mockTime},
|
|
},
|
|
expectMetricsCall: true,
|
|
expectedJobDeploymentStatus: rayv1.JobDeploymentStatusComplete,
|
|
expectedRetryCount: 0,
|
|
expectedDuration: 60.0,
|
|
},
|
|
{
|
|
name: "non-terminal to failed state should emit metrics",
|
|
originalRayJobStatus: rayv1.RayJobStatus{
|
|
JobDeploymentStatus: rayv1.JobDeploymentStatusRunning,
|
|
},
|
|
rayJobStatus: rayv1.RayJobStatus{
|
|
JobDeploymentStatus: rayv1.JobDeploymentStatusFailed,
|
|
StartTime: &metav1.Time{Time: mockTime},
|
|
},
|
|
expectMetricsCall: true,
|
|
expectedJobDeploymentStatus: rayv1.JobDeploymentStatusFailed,
|
|
expectedRetryCount: 0,
|
|
expectedDuration: 60.0,
|
|
},
|
|
{
|
|
name: "non-terminal to retrying state should emit metrics",
|
|
originalRayJobStatus: rayv1.RayJobStatus{
|
|
JobDeploymentStatus: rayv1.JobDeploymentStatusRunning,
|
|
Failed: ptr.To[int32](2),
|
|
},
|
|
rayJobStatus: rayv1.RayJobStatus{
|
|
JobDeploymentStatus: rayv1.JobDeploymentStatusRetrying,
|
|
StartTime: &metav1.Time{Time: mockTime},
|
|
},
|
|
expectMetricsCall: true,
|
|
expectedJobDeploymentStatus: rayv1.JobDeploymentStatusRetrying,
|
|
expectedRetryCount: 2,
|
|
expectedDuration: 60.0,
|
|
},
|
|
{
|
|
name: "non-terminal to non-terminal state should not emit metrics",
|
|
originalRayJobStatus: rayv1.RayJobStatus{
|
|
JobDeploymentStatus: rayv1.JobDeploymentStatusInitializing,
|
|
},
|
|
rayJobStatus: rayv1.RayJobStatus{
|
|
JobDeploymentStatus: rayv1.JobDeploymentStatusRunning,
|
|
},
|
|
expectMetricsCall: false,
|
|
},
|
|
}
|
|
|
|
for _, tt := range tests {
|
|
t.Run(tt.name, func(t *testing.T) {
|
|
ctrl := gomock.NewController(t)
|
|
mockObserver := mocks.NewMockRayJobMetricsObserver(ctrl)
|
|
if tt.expectMetricsCall {
|
|
mockObserver.EXPECT().
|
|
ObserveRayJobExecutionDuration(
|
|
rayJobName,
|
|
rayJobNamespace,
|
|
rayJobUID,
|
|
tt.expectedJobDeploymentStatus,
|
|
tt.expectedRetryCount,
|
|
mock.MatchedBy(func(d float64) bool {
|
|
// Allow some wiggle room in timing
|
|
return math.Abs(d-tt.expectedDuration) < 1.0
|
|
}),
|
|
).Times(1)
|
|
}
|
|
|
|
emitRayJobExecutionDuration(mockObserver, rayJobName, rayJobNamespace, rayJobUID, tt.originalRayJobStatus, tt.rayJobStatus)
|
|
})
|
|
}
|
|
}
|
|
|
|
func TestGetSubmitterTemplate_WithEnableK8sTokenAuth(t *testing.T) {
|
|
rayJob := &rayv1.RayJob{
|
|
Spec: rayv1.RayJobSpec{
|
|
SubmissionMode: rayv1.K8sJobMode,
|
|
},
|
|
}
|
|
rayCluster := &rayv1.RayCluster{
|
|
ObjectMeta: metav1.ObjectMeta{
|
|
Name: "test-raycluster",
|
|
},
|
|
Spec: rayv1.RayClusterSpec{
|
|
AuthOptions: &rayv1.AuthOptions{
|
|
Mode: rayv1.AuthModeToken,
|
|
EnableK8sTokenAuth: ptr.To(true),
|
|
},
|
|
HeadGroupSpec: rayv1.HeadGroupSpec{
|
|
Template: corev1.PodTemplateSpec{
|
|
Spec: corev1.PodSpec{
|
|
Containers: []corev1.Container{
|
|
{
|
|
Image: "rayproject/ray",
|
|
},
|
|
},
|
|
},
|
|
},
|
|
},
|
|
},
|
|
}
|
|
|
|
template, err := getSubmitterTemplate(rayJob, rayCluster)
|
|
require.NoError(t, err)
|
|
|
|
// Check volume
|
|
foundVolume := false
|
|
for _, v := range template.Spec.Volumes {
|
|
if v.Name == utils.RayTokenVolumeName && v.Projected != nil {
|
|
foundVolume = true
|
|
break
|
|
}
|
|
}
|
|
assert.True(t, foundVolume, "Submitter Pod should have the ray-token volume")
|
|
|
|
// Check volume mount
|
|
foundVolumeMount := false
|
|
for _, vm := range template.Spec.Containers[utils.RayContainerIndex].VolumeMounts {
|
|
if vm.Name == utils.RayTokenVolumeName && vm.MountPath == utils.RayTokenMountPath {
|
|
foundVolumeMount = true
|
|
break
|
|
}
|
|
}
|
|
assert.True(t, foundVolumeMount, "Submitter container should have the ray-token volume mount")
|
|
}
|
|
|
|
func TestBatchSchedulerOnCompletionCalledWhenRayJobComplete(t *testing.T) {
|
|
tests := []struct {
|
|
name string
|
|
jobDeploymentStatus rayv1.JobDeploymentStatus
|
|
cleanupDidUpdate bool
|
|
cleanupErr error
|
|
expectCleanupCalled bool
|
|
expectCleanupEvent string
|
|
}{
|
|
{
|
|
name: "Complete status - cleanup performed successfully",
|
|
jobDeploymentStatus: rayv1.JobDeploymentStatusComplete,
|
|
cleanupDidUpdate: true,
|
|
cleanupErr: nil,
|
|
expectCleanupCalled: true,
|
|
expectCleanupEvent: string(utils.BatchSchedulerCleanedUp),
|
|
},
|
|
{
|
|
name: "Complete status - cleanup no-op (no resources to clean)",
|
|
jobDeploymentStatus: rayv1.JobDeploymentStatusComplete,
|
|
cleanupDidUpdate: false,
|
|
cleanupErr: nil,
|
|
expectCleanupCalled: true,
|
|
expectCleanupEvent: "",
|
|
},
|
|
{
|
|
name: "Complete status - cleanup returns error",
|
|
jobDeploymentStatus: rayv1.JobDeploymentStatusComplete,
|
|
cleanupDidUpdate: false,
|
|
cleanupErr: errors.New("cleanup failed"),
|
|
expectCleanupCalled: true,
|
|
expectCleanupEvent: string(utils.FailedToCleanupBatchScheduler),
|
|
},
|
|
{
|
|
name: "Failed status - cleanup performed successfully",
|
|
jobDeploymentStatus: rayv1.JobDeploymentStatusFailed,
|
|
cleanupDidUpdate: true,
|
|
cleanupErr: nil,
|
|
expectCleanupCalled: true,
|
|
expectCleanupEvent: string(utils.BatchSchedulerCleanedUp),
|
|
},
|
|
}
|
|
|
|
for _, tc := range tests {
|
|
t.Run(tc.name, func(t *testing.T) {
|
|
newScheme := runtime.NewScheme()
|
|
_ = rayv1.AddToScheme(newScheme)
|
|
|
|
rayJob := &rayv1.RayJob{
|
|
ObjectMeta: metav1.ObjectMeta{
|
|
Name: "test-rayjob",
|
|
Namespace: "default",
|
|
},
|
|
Spec: rayv1.RayJobSpec{
|
|
Entrypoint: "echo hello",
|
|
RayClusterSpec: &rayv1.RayClusterSpec{
|
|
HeadGroupSpec: rayv1.HeadGroupSpec{
|
|
Template: corev1.PodTemplateSpec{
|
|
Spec: corev1.PodSpec{
|
|
Containers: []corev1.Container{
|
|
{
|
|
Name: "ray-head",
|
|
Image: "rayproject/ray:latest",
|
|
},
|
|
},
|
|
},
|
|
},
|
|
},
|
|
},
|
|
},
|
|
Status: rayv1.RayJobStatus{
|
|
JobDeploymentStatus: tc.jobDeploymentStatus,
|
|
JobStatus: rayv1.JobStatusSucceeded,
|
|
RayClusterName: "test-raycluster",
|
|
},
|
|
}
|
|
|
|
fakeClient := clientFake.NewClientBuilder().
|
|
WithScheme(newScheme).
|
|
WithRuntimeObjects(rayJob).
|
|
WithStatusSubresource(rayJob).
|
|
Build()
|
|
|
|
fakeScheduler := &fakeBatchScheduler{
|
|
cleanupDidUpdate: tc.cleanupDidUpdate,
|
|
cleanupErr: tc.cleanupErr,
|
|
}
|
|
schedulerManager := batchscheduler.NewSchedulerManagerForTest(fakeScheduler)
|
|
|
|
recorder := record.NewFakeRecorder(100)
|
|
|
|
reconciler := &RayJobReconciler{
|
|
Client: fakeClient,
|
|
Recorder: recorder,
|
|
Scheme: newScheme,
|
|
options: RayJobReconcilerOptions{
|
|
BatchSchedulerManager: schedulerManager,
|
|
},
|
|
}
|
|
|
|
ctx := context.Background()
|
|
_, err := reconciler.Reconcile(ctx, reconcile.Request{
|
|
NamespacedName: types.NamespacedName{
|
|
Name: rayJob.Name,
|
|
Namespace: rayJob.Namespace,
|
|
},
|
|
})
|
|
// The Reconcile should not return an error for terminal states
|
|
require.NoError(t, err)
|
|
|
|
// Verify CleanupOnCompletion was called
|
|
assert.True(t, fakeScheduler.cleanupCalled, "CleanupOnCompletion should have been called when RayJob is in %s status", tc.jobDeploymentStatus)
|
|
assert.Equal(t, rayJob.Name, fakeScheduler.cleanupObject.GetName(), "CleanupOnCompletion should receive the correct RayJob object")
|
|
assert.Equal(t, rayJob.Namespace, fakeScheduler.cleanupObject.GetNamespace(), "CleanupOnCompletion should receive the correct RayJob namespace")
|
|
|
|
// Verify the expected event was emitted
|
|
if tc.expectCleanupEvent != "" {
|
|
var foundEvent bool
|
|
for len(recorder.Events) > 0 {
|
|
event := <-recorder.Events
|
|
if strings.Contains(event, tc.expectCleanupEvent) {
|
|
foundEvent = true
|
|
break
|
|
}
|
|
}
|
|
assert.True(t, foundEvent, "Expected event %q to be emitted", tc.expectCleanupEvent)
|
|
}
|
|
})
|
|
}
|
|
}
|
|
|
|
func TestGetJobStatusCheckTimeoutSeconds(t *testing.T) {
|
|
t.Setenv(utils.RAYJOB_STATUS_CHECK_TIMEOUT_SECONDS, "120")
|
|
assert.Equal(t, 120, getJobStatusCheckTimeoutSeconds())
|
|
|
|
t.Setenv(utils.RAYJOB_STATUS_CHECK_TIMEOUT_SECONDS, "invalid")
|
|
assert.Equal(t, utils.DEFAULT_RAYJOB_STATUS_CHECK_TIMEOUT_SECONDS, getJobStatusCheckTimeoutSeconds())
|
|
|
|
os.Unsetenv(utils.RAYJOB_STATUS_CHECK_TIMEOUT_SECONDS)
|
|
assert.Equal(t, utils.DEFAULT_RAYJOB_STATUS_CHECK_TIMEOUT_SECONDS, getJobStatusCheckTimeoutSeconds())
|
|
}
|
|
|
|
func TestRecordJobStatusCheckFailure(t *testing.T) {
|
|
ctx := context.Background()
|
|
rayJob := &rayv1.RayJob{
|
|
Status: rayv1.RayJobStatus{JobId: "test-job"},
|
|
}
|
|
testErr := errors.New("connection refused")
|
|
|
|
needsPersist := recordJobStatusCheckFailure(ctx, rayJob, testErr)
|
|
assert.True(t, needsPersist)
|
|
assert.NotNil(t, rayJob.Status.JobStatusCheckFailureStartTime)
|
|
|
|
needsPersist = recordJobStatusCheckFailure(ctx, rayJob, testErr)
|
|
assert.False(t, needsPersist)
|
|
}
|
|
|
|
func TestCheckJobStatusCheckTimeoutAndUpdateStatusIfNeeded(t *testing.T) {
|
|
ctx := context.Background()
|
|
rayJob := &rayv1.RayJob{
|
|
Status: rayv1.RayJobStatus{JobId: "test-job"},
|
|
}
|
|
|
|
assert.False(t, checkJobStatusCheckTimeoutAndUpdateStatusIfNeeded(ctx, rayJob))
|
|
|
|
rayJob.Status.JobStatusCheckFailureStartTime = &metav1.Time{Time: time.Now()}
|
|
assert.False(t, checkJobStatusCheckTimeoutAndUpdateStatusIfNeeded(ctx, rayJob))
|
|
|
|
t.Setenv(utils.RAYJOB_STATUS_CHECK_TIMEOUT_SECONDS, "1")
|
|
rayJob.Status.JobStatusCheckFailureStartTime = &metav1.Time{Time: time.Now().Add(-2 * time.Second)}
|
|
assert.True(t, checkJobStatusCheckTimeoutAndUpdateStatusIfNeeded(ctx, rayJob))
|
|
assert.Equal(t, rayv1.JobDeploymentStatusFailed, rayJob.Status.JobDeploymentStatus)
|
|
assert.Equal(t, rayv1.JobStatusCheckTimeoutExceeded, rayJob.Status.Reason)
|
|
assert.Contains(t, rayJob.Status.Message, "exceeded timeout of 1s")
|
|
}
|
|
|
|
// TestBatchSchedulerCleanupCalledWhenRayJobSuspending verifies that batch scheduler resources are
|
|
// cleaned up when a RayJob is suspended/retrying. Unlike the terminal (Complete/Failed) path, the
|
|
// suspend path must requeue on cleanup failure rather than swallow the error, otherwise the PodGroup
|
|
// would leak once the status transitions to Suspended (which is no longer requeued).
|
|
func TestBatchSchedulerCleanupCalledWhenRayJobSuspendingOrRetrying(t *testing.T) {
|
|
tests := []struct {
|
|
name string
|
|
jobDeploymentStatus rayv1.JobDeploymentStatus
|
|
expectedStatus rayv1.JobDeploymentStatus
|
|
cleanupErr error
|
|
cleanupDidUpdate bool
|
|
expectReconcileErr bool
|
|
expectCleanupEvent string
|
|
}{
|
|
{
|
|
name: "Suspending - cleanup succeeds, transitions to Suspended",
|
|
jobDeploymentStatus: rayv1.JobDeploymentStatusSuspending,
|
|
expectedStatus: rayv1.JobDeploymentStatusSuspended,
|
|
cleanupDidUpdate: true,
|
|
expectCleanupEvent: string(utils.BatchSchedulerCleanedUp),
|
|
},
|
|
{
|
|
name: "Retrying - cleanup succeeds, transitions to New",
|
|
jobDeploymentStatus: rayv1.JobDeploymentStatusRetrying,
|
|
expectedStatus: rayv1.JobDeploymentStatusNew,
|
|
cleanupDidUpdate: true,
|
|
expectCleanupEvent: string(utils.BatchSchedulerCleanedUp),
|
|
},
|
|
{
|
|
name: "Suspending - cleanup fails, requeues and stays Suspending",
|
|
jobDeploymentStatus: rayv1.JobDeploymentStatusSuspending,
|
|
expectedStatus: rayv1.JobDeploymentStatusSuspending,
|
|
cleanupErr: errors.New("cleanup failed"),
|
|
expectReconcileErr: true,
|
|
expectCleanupEvent: string(utils.FailedToCleanupBatchScheduler),
|
|
},
|
|
{
|
|
name: "Retrying - cleanup fails, requeues and stays Retrying",
|
|
jobDeploymentStatus: rayv1.JobDeploymentStatusRetrying,
|
|
expectedStatus: rayv1.JobDeploymentStatusRetrying,
|
|
cleanupErr: errors.New("cleanup failed"),
|
|
expectReconcileErr: true,
|
|
expectCleanupEvent: string(utils.FailedToCleanupBatchScheduler),
|
|
},
|
|
}
|
|
|
|
for _, tc := range tests {
|
|
t.Run(tc.name, func(t *testing.T) {
|
|
newScheme := runtime.NewScheme()
|
|
_ = rayv1.AddToScheme(newScheme)
|
|
_ = batchv1.AddToScheme(newScheme)
|
|
|
|
rayJob := &rayv1.RayJob{
|
|
ObjectMeta: metav1.ObjectMeta{
|
|
Name: "test-rayjob",
|
|
Namespace: "default",
|
|
},
|
|
Spec: rayv1.RayJobSpec{
|
|
Entrypoint: "echo hello",
|
|
RayClusterSpec: &rayv1.RayClusterSpec{
|
|
HeadGroupSpec: rayv1.HeadGroupSpec{
|
|
Template: corev1.PodTemplateSpec{
|
|
Spec: corev1.PodSpec{
|
|
Containers: []corev1.Container{
|
|
{
|
|
Name: "ray-head",
|
|
Image: "rayproject/ray:latest",
|
|
},
|
|
},
|
|
},
|
|
},
|
|
},
|
|
},
|
|
},
|
|
Status: rayv1.RayJobStatus{
|
|
JobDeploymentStatus: tc.jobDeploymentStatus,
|
|
// The RayCluster and submitter Job are intentionally absent from the fake client,
|
|
// so deleteClusterResources/deleteSubmitterJob treat them as already deleted and
|
|
// the reconcile proceeds to the batch scheduler cleanup.
|
|
RayClusterName: "test-raycluster",
|
|
},
|
|
}
|
|
|
|
fakeClient := clientFake.NewClientBuilder().
|
|
WithScheme(newScheme).
|
|
WithRuntimeObjects(rayJob).
|
|
WithStatusSubresource(rayJob).
|
|
Build()
|
|
|
|
fakeScheduler := &fakeBatchScheduler{
|
|
cleanupDidUpdate: tc.cleanupDidUpdate,
|
|
cleanupErr: tc.cleanupErr,
|
|
}
|
|
schedulerManager := batchscheduler.NewSchedulerManagerForTest(fakeScheduler)
|
|
|
|
recorder := record.NewFakeRecorder(100)
|
|
|
|
reconciler := &RayJobReconciler{
|
|
Client: fakeClient,
|
|
Recorder: recorder,
|
|
Scheme: newScheme,
|
|
options: RayJobReconcilerOptions{
|
|
BatchSchedulerManager: schedulerManager,
|
|
},
|
|
}
|
|
|
|
ctx := context.Background()
|
|
_, err := reconciler.Reconcile(ctx, reconcile.Request{
|
|
NamespacedName: types.NamespacedName{
|
|
Name: rayJob.Name,
|
|
Namespace: rayJob.Namespace,
|
|
},
|
|
})
|
|
if tc.expectReconcileErr {
|
|
require.Error(t, err, "Reconcile should requeue with an error when cleanup fails during suspend/retry")
|
|
} else {
|
|
require.NoError(t, err)
|
|
}
|
|
|
|
require.True(t, fakeScheduler.cleanupCalled, "CleanupOnCompletion should have been called when RayJob is in %s status", tc.jobDeploymentStatus)
|
|
assert.Equal(t, rayJob.Name, fakeScheduler.cleanupObject.GetName(), "CleanupOnCompletion should receive the correct RayJob object")
|
|
|
|
updated := &rayv1.RayJob{}
|
|
require.NoError(t, fakeClient.Get(ctx, types.NamespacedName{Name: rayJob.Name, Namespace: rayJob.Namespace}, updated))
|
|
assert.Equal(t, tc.expectedStatus, updated.Status.JobDeploymentStatus)
|
|
|
|
if tc.expectCleanupEvent != "" {
|
|
var foundEvent bool
|
|
for len(recorder.Events) > 0 {
|
|
event := <-recorder.Events
|
|
if strings.Contains(event, tc.expectCleanupEvent) {
|
|
foundEvent = true
|
|
break
|
|
}
|
|
}
|
|
assert.True(t, foundEvent, "Expected event %q to be emitted", tc.expectCleanupEvent)
|
|
}
|
|
})
|
|
}
|
|
}
|