Hi Folks 👋,
I've a GCP K8 Cluster that has filestore enabled as an addon via this command
gcloud container clusters update XXXXX \
--update-addons=GcpFilestoreCsiDriver=ENABLED \
--zone=us-central1-a \
--project=XXXXX
➜ ~ kubectl get csidrivers
NAME ATTACHREQUIRED PODINFOONMOUNT STORAGECAPACITY TOKENREQUESTS REQUIRESREPUBLISH MODES AGE
filestore.csi.storage.gke.io false true false <unset> false Persistent 7d21h
I've noticed I get this warning during node pool resizing events in particular.
kubernetes.io/csi: attacher.MountDevice failed to create newCsiDriverClient: driver name filestore.csi.storage.gke.io not found in the list of registered CSI drivers
Having looked at other CSI driver implementations it appears that there is a slight race condition where the node boots up for first time but the filestore daemonsets required haven't had a chance yet to start and pods with filestore volumes get assigned to the node prematurely causing the warning. Example from another driver https://github.com/GoogleCloudPlatform/gcs-fuse-csi-driver/blob/main/docs/troubleshooting.md#csi-driver-enablement-issues
This warning isn't the end of the world as the pods eventually go get scheduled but I would like to resolve this noisy log. My idea was possibly when the daemonsets required by filestore are ready a label is added to the node like filestore-ready: true then I can check this label for my pods via node selector. I would then hope this would eliminate the warning log as a result
Hi Folks 👋,
I've a GCP K8 Cluster that has filestore enabled as an addon via this command
I've noticed I get this warning during node pool resizing events in particular.
Having looked at other CSI driver implementations it appears that there is a slight race condition where the node boots up for first time but the filestore daemonsets required haven't had a chance yet to start and pods with filestore volumes get assigned to the node prematurely causing the warning. Example from another driver https://github.com/GoogleCloudPlatform/gcs-fuse-csi-driver/blob/main/docs/troubleshooting.md#csi-driver-enablement-issues
This warning isn't the end of the world as the pods eventually go get scheduled but I would like to resolve this noisy log. My idea was possibly when the daemonsets required by filestore are ready a label is added to the node like
filestore-ready: truethen I can check this label for my pods via node selector. I would then hope this would eliminate the warning log as a result