"H. Peter Anvin" wrote:
>
[ .. ]
> That's yet another problem (semantic.) But the implementation problems
> alone are hideous. Right now autofs triggers a mount on any request for
> a lookup. This is nice, easy, and clean. However, if you want to do
> mount filtering based on the operation that triggered a lookup, not only
> do you need to worry about knowing what operation triggered the lookup,
> you *ALSO* have the problem that you won't *get* the lookup the next
> time anything comes looking for this directory, and it might be
> something for which you really need to mount.
>
> This gets very ugly very fast, and is probably undoable without VFS
> changes, which would impose overhead for every operation for every
> filesystem.
>
Hmm. Speaking as an unexperienced occasional hacker, I don't think we
need to delve into that complexity. Thing is, we probably can get away
with a two-stage scenario:
1.) We do a lookup (a la nmblookup) of all the available _servers_ in
the network. Those are not mounted, but a simple mkdir for each of them
will suffice.
2.) On each access of one of those _directories_ we mount all available
shares for that server under this directory.
Sounds feasible and should be easily done by some autofs modules. Almost
too easy. Do I miss something fundamentally here ?
Cheers,
Hannes
--
Hannes Reinecke <[EMAIL PROTECTED]>
Fluid Loading and Instrumentation Center Tel: (+44) 131 451 3149
Dept. of Civil & Offshore Engineering Fax: (+44) 131 451 3154
Heriot Watt University, Edinburgh EH14 4AS
