[DRAFT] poc seamless worker update

[itechdima]

Problem

Solution

Testing

Release Notes

[Copilot]

Pull request overview

Adds a NodeSet-level update strategy knob for worker (OpenKruise) StatefulSets and introduces a “self-restart” mechanism for workers by deleting the running pod via the Kubernetes API, with corresponding CRD/schema and RBAC updates.

Changes:

• Add spec.updateStrategy to NodeSet API/CRDs and plumb it through values/rendering.
• Render Kruise StatefulSet UpdateStrategy / VolumeClaimUpdateStrategy based on the configured strategy.
• Change worker reboot script to delete its own pod, and grant worker Role permission to delete pods.

Reviewed changes

Copilot reviewed 9 out of 9 changed files in this pull request and generated 9 comments.

Show a summary per file

File	Description
internal/values/slurm_nodeset.go	Plumbs UpdateStrategy from NodeSet spec into rendered values.
internal/render/worker/statefulset.go	Switches StatefulSet strategy rendering to be configurable via UpdateStrategy.
internal/render/worker/role.go	Expands worker RBAC to allow pod deletion (needed by reboot script).
internal/consts/statefulset.go	Introduces UpdateStrategy enum constants.
images/common/scripts/reboot.sh	Replaces host reboot with Kubernetes API pod deletion.
api/v1alpha1/nodeset_types.go	Adds spec.updateStrategy to the NodeSet API type with validation/default.
config/crd/bases/slurm.nebius.ai_nodesets.yaml	CRD schema update for updateStrategy (incl. required/default/enum).
helm/soperator/crds/slurmcluster-crd.yaml	Helm-packaged CRD schema update for updateStrategy.
helm/soperator-crds/templates/slurmcluster-crd.yaml	Helm template CRD schema update for updateStrategy.

---

💡 Add Copilot custom instructions for smarter, more guided reviews. Learn how to get started.