Delete a node | NVIDIA Kubernetes Engine

Deletes a single node from a running cluster. Include a machineHealthIssue to flag the underlying machine for repair.

For control-plane nodes, a delete sets deleteRequested; NVKE then starts teardowns one node at a time. forceUnreachable is applied automatically when the node is unhealthy.

Also returns 202, with no further effect, if the node, pool or cluster is already being deleted or a delete is already requested.

Returns 400 when deletion would leave no joined control-plane node or machineHealthIssue is invalid. Delete the cluster to remove its last joined control-plane node. Deleting a control-plane node requires controlPlaneVIP because the controller replaces it. Worker node deletion is unaffected.

Authentication

AuthorizationBearer

A JWT from an issuer the Operator has configured, carried as Authorization: Bearer <token>.

Path parameters

nodeIDstringRequiredformat: "uuid"

Node ID.

Request

Optional. Flags the machine for repair, or tolerates an unreachable node.

machineHealthIssueobjectOptional

Flags the underlying machine for repair.

forceUnreachablebooleanOptionalDefaults to false

Treat the node as unreachable: graceful drain and shutdown are attempted, and failures are tolerated. Failures removing a control-plane node’s etcd member are not tolerated.

Response

Deletion started or, for control-plane nodes, recorded

Errors

400

Bad Request Error

401

Unauthorized Error

403

Forbidden Error

404

Not Found Error