Trident PVC mount fails with Autogrow and Concurrency

• NetApp Trident 26.06.1
• ONTAP 9.18.1P6 (Cluster-Mode)
• Kubernetes v1.35.5 (OpenShift 4.22.x)
• StorageClass: ontap-nvme-block (autogrowPolicy enabled)

Pods in the openshift-monitoring namespace fail to start and remain stuck in CreateContainerConfigError due to repeated errors mounting PVCs backed by ONTAP via Trident.

Observed symptoms:

• Multiple NetApp Trident and ONTAP alerts:
  volume "trident_pvc_<uuid>" is full (using or reserving 91% of space and 0% of inodes)
• On the ONTAP system, the volume/filesystem appears under threshold (not full).
• Kubernetes pod events (example):

  Warning  Failed  kubelet  Error: failed to prepare subPath for volumeMount "<namespace>-k8s-db" of container

• Kubelet logs:

  error resolving symlinks in "/var/lib/kubelet/pods/.../volumes/kubernetes.io~csi/pvc-<uuid>/mount": lstat ...: input/output error

• Pod status:

  <namespace>-k8s-0   5/6  CreateContainerConfigError
  <namespace>-k8s-1   5/6  CreateContainerConfigError

• Trident node logs (decisive error):

  fatal error: concurrent map writesgithub.com/netapp/trident/frontend/csi/node_helpers/kubernetes.(*helper).AddPublishedPath/go/src/github.com/netapp/trident/frontend/csi/node_helpers/kubernetes/plugin.go:214

• Repeated warnings/errors in Trident node logs:
  • Failed to add volume(tvp) to effectivePolicyCache ... already exists in cache
  • Failed to add volume(tvp) to assorter ... already exists in assorter
  • ProcessedCount=0 ... errorCount=N ... during autogrow reconciliation
• Affected PVCs:
  • <namespace>-k8s-db-prometheus-k8s-0 (pvc-name)
  • <namespace>-k8s-db-prometheus-k8s-1 (pvc-name)
• Red Hat support confirmed issue is isolated to storage backend and not a permissions/configuration issue on the Kubernetes side.