Node reboot can cause lingering paths in /var/lib/kubelet/pods/x-x-x-x-x/volumes/greenhost~cloud/
We have some occurrences of this problem in the kubelet logs:
jun 24 09:42:04 worker1 kubelet[890]: E0624 09:42:04.261215 890 kubelet_volumes.go:154] orphaned pod "251a1af9-17c4-4441-9507-2ff003492daf" found, but volume paths are still present on disk : There were a total of 10 errors similar to this. Turn up verbosity to see them.
From this comment I presume that the paths are there because a shutdown caused the volume to be unmounted without properly removing the data from the /var/lib/kubelet/pods/x-x-x-x-x/volumes/greenhost~cloud/
directory
From the same issue we can conclude that a workaround is to manually remove the directories that were created for greenhost~cloud
based PVCs. This should be safe as long as they are empty.
The empty lingering directories will be automatically removed in a newer version of Kubernetes (1.21, it seems)