Concourse on K8s: Volumes keep disappearing

I deployed Concourse 5.3.0 on a K8s cluster on GCP using the Helm chart of version 6.5.0 from and keep getting the following error in my resources and sometimes it interrupts the execution of tasks, too:

volume 'ce10f092-6d2e-4c91-6b4d-eb0dfee4a201' disappeared from worker 'ci-worker-4'

Of course, the volume UUID differs from message to message.

Any ideas why this is happening?