On Fri, 2021-10-22 at 22:07 -0700, Jeremy Allison via samba
wrote:> On Sat, Oct 23, 2021 at 12:03:18AM -0400, Eric Levy via samba wrote:
> > In my earlier conversation in this group, I described my needs as
> > follows:
> >
> > What I want is multiple users on the client accessing the same
> > mount
> > but with different permissions enforced for each. I want the
> > permissions to reflect the permissions for the corresponding
> > users
> > on the NAS.
> >
> > It seems by now it has been made clear that it is impossible to
> > achieve this result without some kind of domain server...
>
> Isn't that the bog-standard standalone file server case,
> with user names on the client mapped into the same user
> names on the server ?
>
> The clients can easily do multi-user mounts, both Windows
> and Linux.
>
> I guess I don't understand exactly what you are asking
> for here.
>
> In your scenario, where are the "users" defined ? How
> does a client have multiple users logged in ? Are
> these local users defined on the client ?
When I inquired earlier to this group, it was explained that multiuser
mounts depend on a domain server, and this explanation is also given in
the documentation. I think the standard standalone case is that all
files in the mount share the same owner viewed by the client, perhaps
with some added support for special users such as "nobody". A mount
that shows different files owned by various regular users is not
supported. The reason is as you say, some mechanism is required to
support a user mapping, which currently is handled only by a domain
server.
The immediate need that interests me personally is that the same set of
users are defined in the client and file server, with association
occurring by equivalence of names, that is, symbolic user identifiers.
The proposed class (3) is much more general, deliberately so, as its
definition is to include all cases not in the other two classes. What
support might be available for various cases in an actual design is
obviously too detailed to discuss in this conversation. The purpose of
this conversation is to explain my view that the cases currently
supported are limiting, according to my understanding and in my view of
what is useful to me personally and presumably also more generally for
many others in a very small but still multiuser environment.