We are working with NAS Snapshots on FAS2552, OnTap 8.3.2 .
Once the Volume snapshots are taken the Snapshots are NFS mounted on a host from where the contents are copied to a secondary storage. At this point we are observing that the inode number of files are changing in different snapshots. For e.g., a file F1 under snapshot 1 is different than the one under snapshot 2. Interestingly this file is not touched at all between the snapshots.
More so, this behavior is observed only when the snapshot is mounted on NFS v3.
See the below outputs -
----- NFS Version 3 --------
10.xx.xx.147:/demo_nas_vol1 on /ver3_mnt type nfs (rw,relatime,vers=3,rsize=65536,wsize=65536,....addr=10.xx.xx.147)
[root@nfshost ~]# stat -c %i /ver3_mnt/.snapshot/snap1_for_inodeTest/fileone.txt 858188230 [root@nfshost ~]# stat -c %i /ver3_mnt/.snapshot/snap2_for_inodeTest/fileone.txt 858188486
---- NFS Version 4 ----- 10.xx.xx.147:/demo_nas_vol1 on /ver4_mnt type nfs4 (rw,relatime,vers=4.0,rsize=65536,wsize=65536,....addr=10.xx.xx.147)
[root@nfshost ~]# stat -c %i /ver4_mnt/.snapshot/snap1_for_inodeTest/fileone.txt 13510 [root@nfshost ~]# stat -c %i /ver4_mnt/.snapshot/snap2_for_inodeTest/fileone.txt 13510
This will still not resolve the Inode issue I am observing while calling stat at the client end, because essentially, for NFSv3, FSID of a file is returned and this FSID is a combination of inode, volume identifier and Snapshot IDs.
However for NFSv4 if v4-fsid-change is enabled, the inode number is returned as FSID instead of FSID.