Kubernetes Worker VM Stuck in Provisioning State

Problem

• Kubernetes cluster is stuck in ScalingUp state & Machine Status is stuck in Provisioning state.

Environment

• Private Cloud Director Kubernetes - v2025.4 and Higher

• Component - Kubernetes

Cause

• Review the machinedeployments from the <SHORTNAME_DEFAULT_SERVICE> namespace. There are no Ready replicas.

$ kubectl get machinedeployments -n <SHORTNAME_DEFAULT_SERVICE>
NAME                             CLUSTER                 REPLICAS   READY   UPDATED   UNAVAILABLE   PHASE       AGE    VERSION
test-no-ssh-3b377fe3             test-no-ssh             1                  1         1             ScalingUp   22m    v1.32.3

• Further checking the machines object, the VM is in failed state. In the yaml output of the machine resource, it complains about a missing key.

$ kubectl get machines -n <SHORTNAME_DEFAULT_SERVICE>
NAME                            CLUSTER   NODENAME   PROVIDERID    PHASE     AGE    VERSION
test-no-ssh-3b377fe3-9xft6-d4h6d test-no-ssh                                                                                                              Failed    25m    v1.32.3

Message:               error creating Openstack instance: Bad request with: [POST https:__example1.platform9.com_nova_v2.1_<id>_servers], error message: {"badRequest": {"code": 400, "message": "Invalid key_name provided."}}
Reason:                InstanceCreateFailed
Severity:              Error
Status:                False

Resolution

• As a prerequisite, public SSH key needs to be added in the PCD Virtualization before creating a Kubernetes cluster. The key name has to be default only, any other name must not be used. This key is used to bootstrapped into the worker nodes and will be used to access the worker nodes later.

• This can be created via UI Networks and Security SSH Keys.

• All users need to have individual SSH Key created.