I think I found the reason of timeouts when NFS volume is mounted: https://blog.devgenius.io/when-k8s-pods-are-stuck-mounting-large-volumes-2915e6656cb8

template:
  spec:
    containers:
      ...
    securityContext:
      fsGroup: 10001
      runAsGroup: 10001
      runAsNonRoot: true
      runAsUser: 10001
      fsGroupChangePolicy: "OnRootMismatch"

With OnRootMismatch the pod doesn't try to change all files owner in the mounted folder.

To support this, I need to update SparkApplication CRD.

This change is tested in ml-stage (by manual editing of CRD) for the streaming data ingestion jobs that were today very slow to start because of the mounting timeouts.