Mount errors due to incorrect disk size formatting.

[hajiler]

A race condition can cause a new Persistent Disk to be formatted with an incorrect filesystem size when a node rapidly attaches and detaches multiple disks. This issue arises when a new disk reuses the same device node path (e.g., /dev/nvme0n5) as a recently detached disk.

A timing gap occurs because the udev service quickly updates the symlinks to point to the new disk's name, while the kernel's block device layer takes longer—sometimes over a second—to recognize that the disk's size has also changed. Consequently, Kubelet initiates the formatting process based on the updated symlink but reads the cached size of the previous disk from the kernel. This results in the filesystem being created with the wrong size, rendering the newly attached disk unusable.