Breaking change with upgrade #612
Description
I have a deployment which has been using the GCSFuse CSI driver with a service account for a while.
With the latest node upgrade to 1.32.3-gke.1927009, the deployment failed with the following error:
Warning FailedMount kubelet MountVolume.SetUp failed for volume "gcs-fuse-csi-ephemeral" : rpc error: code = FailedPrecondition desc = Workload Identity Federation is not enabled on node. Please make sure this is enabled on both cluster and node pool level (https://cloud.google.com/kubernetes-engine/docs/how-to/workload-identity)
After I enabled GKE Metadata Server on the pool, it started working again.
The init container imageID is:
gke.gcr.io/gcs-fuse-csi-driver-sidecar-mounter@sha256:16d64badf2336967c868cf71ccd5d405cbb4703d5438e0edab1403f61008c49c
Does the driver no longer support a service account? This kind of breaking change should not take place as part of a routine node upgrade.