ONTAP9.6P3 /.snapshot automount

[vernal niche]

A customer using ONTAP 9.6P3 is using NFS service, and one PATH is MOUNTED on three servers and used at the same time.

If you have three servers, A, B, and C, it is confirmed that the path './snapshot' is mounted when you do 'df-h' only on server B.

Customers are uncomfortable with this, so why is '.'/snapshot' automatically mounted on only one of the three hosts?

If I hide the snap directory through the command vserver nfs modify -vserver <vserver> -v3-hide-snapshot {enabled|disabled}, won't it be difficult for customers to access the ./snapshot path and cause file recovery problems?

I'm curious about the proper workaround!

Summary) Only on one of several servers /.Snapshot mounts automatically, customer experiences inconvenience.

[normal aspen]

Modern Linux systems automatically mount .snapshot seperatly when accessed, because ONTAP presents a different fsid. If no process keeps the mount open, it will automatically unmount. I think this KB article will help to understand and show a workaround: https://kb.netapp.com/onprem/ontap/da/NAS/A_Linux_system_running_kernel_2.6x_creates_a_mount_point_when_accessing_the_.snapshot_directory_via_Network_File_System

[vernal niche]

Yeah I know it's automatically unmounted, but it's uncomfortable for customers to use it and be seen only on certain servers.

[normal aspen]

Check the KB, you can reconfigure ONTAP to avoid the fsid change.

[vernal niche]

I don't understand well, but I'll read the thank you document carefully.

[river flame]

this is an NFSv4 thing. If you use NFSv3 this shouldn't happen

because it's a different "filesystem" on the server (it's a snapshot but a snapshot has a different file system ID, otherwise you would get duplicate handles on a file and all of its versions), this will propagate to the client, to make it explicit that you're crossing FS boundaries (which impacts df . for example)