Is there a way to make a browse-able automount directory without mounting the subdirectories?
I am looking at moving from statically mounted NFS shares, to dynamically mounted NFS shares. The shares are hosted on OS X and OpenSolaris systems, the clients are running Debian and Ubuntu.
I would like the user directories in /home to be dynamically mounted so that the OpenSolaris shares can each be in its own ZFS partition for quota, compression, and duplication tweaking. The OpenSolaris/ZFS best practices guide suggests mounting each directory individually and using an automounter. There is also a danger with a lot of mounted NFS shares running into issues.
My users to need to regularly copy data from another user's home directory, and I think they will panic if the list of user directories is not browse-able.
This illustrates the point, but there must be better solution by now. When I set this up, the directory that hosts the to-be-mounted directory starts empty. It is only after attempting a function (like 'ls folder_name') on a directory I know should exist, that the remote directory is mounted. I understand this is normal for autofs and amd.
If I make symbolic links to the directories so that I have a browse-able tree for my users, each remote share is mounted automatically when the directory that hosts the symbolic links is 'ls'ed. This is largely a waste of mounting as most of those mounts will not be used.