Skip to content

feat(operator): Add validation for required containers in replicatedJobs #2722

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
merged 4 commits into from
Oct 20, 2025
+72 −10
Merged

feat(operator): Add validation for required containers in replicatedJobs #2722

Show file tree
Hide file tree
Changes from all commits
Commits
Show all changes
4 commits
Select commit Hold shift + click to select a range
cff6790
feat(webhook): Add validation for required containers in replicatedJobs.
Electronic-Waste Jul 8, 2025
0efd0f8
test(webhook): Add UTs for validation in required containers.
Electronic-Waste Jul 8, 2025
6f849f0
fix(lint): fix lint error.
Electronic-Waste Jul 8, 2025
0797125
fix(webhook): add global map & remove launcher check.
Electronic-Waste Oct 18, 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
20 changes: 20 additions & 0 deletions pkg/util/testing/wrapper.go
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -270,6 +270,26 @@ func (j *JobSetWrapper) Container(rJobName, containerName, image string, command
return j
}

func (j *JobSetWrapper) ReplaceContainer(rJobName, containerName, newContainerName, image string, command, args []string, res corev1.ResourceList) *JobSetWrapper {
for i, rJob := range j.Spec.ReplicatedJobs {
if rJob.Name == rJobName {
for k, container := range rJob.Template.Spec.Template.Spec.Containers {
if container.Name == containerName {
j.Spec.ReplicatedJobs[i].Template.Spec.Template.Spec.Containers[k] = corev1.Container{
Name: newContainerName,
Image: image,
Command: command,
Args: args,
Resources: corev1.ResourceRequirements{Requests: res},
}
return j
}
}
}
}
return j
}

func (j *JobSetWrapper) ContainerTrainerPorts(ports []corev1.ContainerPort) *JobSetWrapper {
for i, rJob := range j.Spec.ReplicatedJobs {
if rJob.Name == constants.Node {
Expand Down
43 changes: 35 additions & 8 deletions pkg/webhooks/trainingruntime_webhook.go
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -18,6 +18,7 @@ package webhooks

import (
"context"
"fmt"

apiruntime "k8s.io/apimachinery/pkg/runtime"
"k8s.io/apimachinery/pkg/util/sets"
Expand All @@ -34,7 +35,16 @@ import (
)

const (
rJobReplicasErrorMsg = "always must be 1"
rJobReplicasErrorMsg = "always must be 1"
rJobContainerNamesErrorMsg = "must contain the required container for the ancestor: %s"
)

var (
expectedContainerNames = map[string]string{
constants.AncestorTrainer: constants.Node,
constants.ModelInitializer: constants.ModelInitializer,
constants.DatasetInitializer: constants.DatasetInitializer,
}
)

type TrainingRuntimeWebhook struct {
Expand Down Expand Up @@ -67,18 +77,35 @@ func validateReplicatedJobs(rJobs []jobsetv1alpha2.ReplicatedJob) field.ErrorLis
Child("replicatedJobs")
var allErrs field.ErrorList
for idx, rJob := range rJobs {
if rJob.Name == constants.Launcher && rJob.Replicas != 1 {
allErrs = append(allErrs, field.Invalid(rJobsPath.Index(idx).Child("replicas"), rJob.Replicas, rJobReplicasErrorMsg))
}

if rJob.Template.Labels == nil {
continue
}

if labelAncestor, ok := rJob.Template.Labels[constants.LabelTrainJobAncestor]; ok && ancestors.Has(labelAncestor) && rJob.Replicas != 1 {
allErrs = append(allErrs, field.Invalid(rJobsPath.Index(idx).Child("replicas"), rJob.Replicas, rJobReplicasErrorMsg))
if labelAncestor, ok := rJob.Template.Labels[constants.LabelTrainJobAncestor]; ok && ancestors.Has(labelAncestor) {
if rJob.Replicas != 1 {
allErrs = append(allErrs, field.Invalid(rJobsPath.Index(idx).Child("replicas"), rJob.Replicas, rJobReplicasErrorMsg))
}

// Validate replicated job contains the required containers.
// Mapping of the ancestor labels to the containers:
// 1. dataset-initializer - dataset-initializer
// 2. model-initializer - model-initializer
// 3. trainer - node
hasRequiredContainer := false
for _, container := range rJob.Template.Spec.Template.Spec.Containers {
Copy link
Contributor

Choose a reason for hiding this comment

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

I wonder how difficult that would be to implement this validation in CEL?

Copy link
Member Author

@Electronic-Waste Electronic-Waste Oct 18, 2025 •
edited
Loading

Choose a reason for hiding this comment

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

Hmm. These fields are defined in JobSetSpec:

trainer/pkg/apis/trainer/v1alpha1/trainingruntime_types.go

Lines 132 to 134 in 467322a

// spec of the desired JobSet which will be created from TrainJob.
// +optional
Spec jobsetv1alpha2.JobSetSpec `json:"spec,omitempty"`

I'm not sure whether can we reference the container name in it (detecting corresponding label) and verify its value. In the contrast, I think webhook is more convenient and do not need to change the API by adding something like:

trainer/pkg/apis/trainer/v1alpha1/trainingruntime_types.go

Line 206 in 467322a

// +kubebuilder:validation:XValidation:rule="self > 0 || self in ['auto', 'cpu', 'gpu']", message="NumProcPerNode must be equal to auto, cpu, gpu, or int value"

if container.Name == expectedContainerNames[labelAncestor] {
hasRequiredContainer = true
break
}
}
if !hasRequiredContainer {
allErrs = append(allErrs, field.Invalid(
rJobsPath.Index(idx).Child("template").Child("spec").Child("template").Child("spec").Child("containers"),
rJob.Template.Spec.Template.Spec.Containers,
fmt.Sprintf(rJobContainerNamesErrorMsg, labelAncestor),
))
}
}

}
return allErrs
}
Expand Down
19 changes: 17 additions & 2 deletions pkg/webhooks/trainingruntime_webhook_test.go
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -21,6 +21,7 @@ import (

"github.com/google/go-cmp/cmp"
"github.com/google/go-cmp/cmp/cmpopts"
corev1 "k8s.io/api/core/v1"
"k8s.io/apimachinery/pkg/util/validation/field"
jobsetv1alpha2 "sigs.k8s.io/jobset/api/jobset/v1alpha2"

Expand Down Expand Up @@ -55,12 +56,26 @@ func TestValidateReplicatedJobs(t *testing.T) {
"2", ""),
field.Invalid(field.NewPath("spec").Child("template").Child("spec").Child("replicatedJobs").Index(1).Child("replicas"),
"2", ""),
field.Invalid(field.NewPath("spec").Child("template").Child("spec").Child("replicatedJobs").Index(2).Child("replicas"),
"2", ""),
field.Invalid(field.NewPath("spec").Child("template").Child("spec").Child("replicatedJobs").Index(3).Child("replicas"),
"2", ""),
},
},
"missing required container in replicatedJobs": {
rJobs: testingutil.MakeJobSetWrapper("ns", "valid").
Replicas(1, constants.Launcher, constants.Node, constants.DatasetInitializer, constants.ModelInitializer).
ReplaceContainer(constants.DatasetInitializer, constants.DatasetInitializer, "test", "", []string{}, []string{}, corev1.ResourceList{}).
ReplaceContainer(constants.ModelInitializer, constants.ModelInitializer, "test", "", []string{}, []string{}, corev1.ResourceList{}).
ReplaceContainer(constants.Node, constants.Node, "test", "", []string{}, []string{}, corev1.ResourceList{}).
Obj().Spec.ReplicatedJobs,
wantError: field.ErrorList{
field.Invalid(field.NewPath("spec").Child("template").Child("spec").Child("replicatedJobs").Index(0).Child("template").Child("spec").Child("template").Child("spec").Child("containers"),
[]corev1.Container{{Name: "test"}}, ""),
field.Invalid(field.NewPath("spec").Child("template").Child("spec").Child("replicatedJobs").Index(1).Child("template").Child("spec").Child("template").Child("spec").Child("containers"),
[]corev1.Container{{Name: "test"}}, ""),
field.Invalid(field.NewPath("spec").Child("template").Child("spec").Child("replicatedJobs").Index(2).Child("template").Child("spec").Child("template").Child("spec").Child("containers"),
[]corev1.Container{{Name: "test"}}, ""),
},
},
}
for name, tc := range cases {
t.Run(name, func(t *testing.T) {
Expand Down
Loading