Skip to content

[Bug] Sidecar mode shouldn't restart head pod when head pod is deleted #4141

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 10 commits into from
Oct 30, 2025
Merged

[Bug] Sidecar mode shouldn't restart head pod when head pod is deleted #4141

Show file tree
Hide file tree
Changes from 4 commits
Commits
Show all changes
10 commits
Select commit Hold shift + click to select a range
c753e82
[Bug] Sidecar mode shouldn't restart head pod when head pod is deleted
400Ping Oct 24, 2025
8304ed3
[Fix] Fix e2e error
400Ping Oct 24, 2025
3741997
[Fix] fix according to rueian's comment
400Ping Oct 25, 2025
9d5b968
[Chore] fix ci error
400Ping Oct 25, 2025
f2ec21a
Update ray-operator/controllers/ray/raycluster_controller.go
400Ping Oct 29, 2025
08cd600
Update ray-operator/controllers/ray/rayjob_controller.go
400Ping Oct 29, 2025
95b21b3
update
Future-Outlier Oct 29, 2025
22b148e
Merge branch 'master' into bug/sidecar-mode-head-pod
Future-Outlier Oct 29, 2025
b7448cf
update
Future-Outlier Oct 29, 2025
a8d2d8e
Trigger CI
Future-Outlier Oct 29, 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
8 changes: 8 additions & 0 deletions ray-operator/controllers/ray/raycluster_controller.go
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -619,6 +619,14 @@ func (r *RayClusterReconciler) reconcilePods(ctx context.Context, instance *rayv
return errstd.New(reason)
}
} else if len(headPods.Items) == 0 {
originatedFrom := utils.GetCRDType(instance.Labels[utils.RayOriginatedFromCRDLabelKey])
if originatedFrom == utils.RayJobCRD {
logger.Info(
"reconcilePods: Found 0 head Pods for a RayJob-managed RayCluster; skipping head creation to let RayJob controller handle the failure",
Copy link
Collaborator

Choose a reason for hiding this comment

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

Won't this cause no head pod to be created at all? We still need to create the first head pod. I think you can check the RayClusterProvisioned condition to decide whether to create one or not.

"rayCluster", instance.Name,
)
return nil
}
// Create head Pod if it does not exist.
logger.Info("reconcilePods: Found 0 head Pods; creating a head Pod for the RayCluster.")
if err := r.createHeadPod(ctx, *instance); err != nil {
Expand Down
17 changes: 17 additions & 0 deletions ray-operator/controllers/ray/rayjob_controller.go
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -276,6 +276,23 @@ func (r *RayJobReconciler) Reconcile(ctx context.Context, request ctrl.Request)
}
}

if rayJobInstance.Spec.SubmissionMode != rayv1.SidecarMode &&
rayJobInstance.Status.JobStatus == rayv1.JobStatusRunning &&
finishedAt != nil {
headPod, err := common.GetRayClusterHeadPod(ctx, r.Client, rayClusterInstance)
if err != nil {
logger.Error(err, "Failed to get head pod for RayCluster")
return ctrl.Result{RequeueAfter: RayJobDefaultRequeueDuration}, err
}
if headPod == nil {
rayJobInstance.Status.JobDeploymentStatus = rayv1.JobDeploymentStatusFailed
rayJobInstance.Status.JobStatus = rayv1.JobStatusFailed
rayJobInstance.Status.Reason = rayv1.AppFailed
rayJobInstance.Status.Message = "Submitter completed but Ray job not found in RayCluster."
break
}
}

// Check the current status of ray jobs
rayDashboardClient, err := r.dashboardClientFunc(rayClusterInstance, rayJobInstance.Status.DashboardURL)
if err != nil {
Expand Down
Loading