Access to Snapshot copies over NFS

On an NFS client, the user can obtain access to Snapshot copies.

The following illustration shows the directory structure on an NFS client with the vol0 volume named toaster mounted on the /n/toaster directory:

In this example, the user can obtain access to Snapshot copies in the /n/toaster/.snapshot directory. Notice that the .snapshot directory is shown only at the mountpoint, although it actually exists in every directory in the tree. The user, however, can only see the .snapshot directory at the mountpoint. That is, the .snapshot directory is accessible by name in each directory, but is only seen in the output of the ls command at the mountpoint.

For example, at the mountpoint of a file system, a directory listing looks like this:

systemA> ls -a
. .. .snapshot dir1 dir2

The same command entered in a directory below the mountpoint does not show the .snapshot directory; for example:

systemA> cd dir1
systemA> ls -a
. .. file1 file2

If you enter the ls command with the directory name .snapshot, you can see a directory for each of the Snapshot copies for the dir1 directory:

systemA> ls .snapshot

hourly.0 hourly.4 nightly.0 nightly.4
hourly.1 hourly.5 nightly.1 nightly.5
hourly.2 hourly.6 nightly.2 weekly.0
hourly.3 hourly.7 nightly.3 weekly.1

If the .snapshot directory entry appeared in every directory, it would cause many commands to work improperly. For instance, all recursive commands for deleting files would fail because everything below the .snapshot directory is read-only. The recursive commands would copy everything in the Snapshot copies as well as files in the active file system. A find command would generate a list much longer than expected.