Deleting an Enterprise PKS cluster may not complete when the vSphere CSI is used to provision persistent volumes
book
Article ID: 316991
calendar_today
Updated On:
Products
VMware Cloud PKS
Issue/Introduction
Symptoms:
A cluster is created in Enterprise PKS 1.7 and vSphere CSI is configured to allow for persistent volumes to be created via vSphere Cloud Native Storage (CNS).
A workload is deployed with persistent volumes that makes use of vSphere CNS.
An attempt to delete the cluster is made but it hangs indefinitely.
You see that the drain-cluster errand task is not completing. You see output similar to the following when running bosh task <id>:
Task 137 | 13:17:33 | Preparing deployment: Preparing deployment Task 137 | 13:17:33 | Deprecation: Global 'properties' are deprecated. Please define 'properties' at the job level. Task 137 | 13:17:36 | Warning: Executing errand on multiple instances in parallel. Use the `--instance` flag to run the errand on a single instance. Task 137 | 13:17:36 | Preparing deployment: Preparing deployment (00:00:03) Task 137 | 13:17:36 | Running errand: worker/64b96419-83bb-48eb-a1c5-5823855df887 (1) Task 137 | 13:17:36 | Running errand: worker/f85cef70-0dc8-41ba-9a47-5c64c844411b (0) Task 137 | 13:17:40 | Running errand: worker/64b96419-83bb-48eb-a1c5-5823855df887 (1) (00:00:04) Task 137 | 13:17:40 | Fetching logs for worker/64b96419-83bb-48eb-a1c5-5823855df887 (1): Finding and packing log files (00:00:01)
Note: The preceding log excerpts are only examples. Date, time, and environmental variables may vary depending on your environment.
Environment
VMware Cloud Native Storage 1.x VMware PKS 1.7.x
Resolution
This is a known issue affecting Enterprise PKS 1.7 when vSphere Cloud Native storage is used. There is currently no resolution.
Workaround: To workaround this issue you can issue the bosh cancel-task <id> command to cancel the drain-cluster errand task.
To prevent this issue from occurring you can manually delete the workload that is using CNS to provision persistent volumes prior to deleting the cluster.