Roy Stogner wrote:
>
> I've been playing with autofs and smbfs for the past couple days, with
> the following discoveries:
>
> HPA is right; smbfs has problems, and mount_smbfs.c is a waste.
> He (and anyone else working on autofs) has better things to do.
>
> The more recent samba versions are getting better client-side support,
> even if samba people pretty much disown smbfs as Linux-only. If
> you're not using the latest samba clients, you should be.
>
> The best solution for smbfs mounting I've seen is the "mount.smb"
> script (and "mount.smbfs" symlink) included in Red Hat's distribution
> of the samba clients. /bin/mount calls this script, and it basically
> lets you "mount -t smb //server/share /mount/point" as if it were
> normal, and the call to smbmount occurs behind the scenes. This
> doesn't affect the mount() system call, but it works for programs like
> autofs's mount_generic.so module that fork/exec the mount command.
>
> I personally think mount_smbfs.c should be thrown away (would this be
> a mount_smbfs change that HPA would be happy with?), smbfs should be
> removed from the list of non-generic filesystems, and people and
> distributions should be encouraged to just make their mount programs
> work right.
>
Oh, this is the only reasonable solution to this problem. I'd be happy
to get rid of mount_smbfs and tell people to get mount.smbfs, as soon as
one appears.
Note the correct filesystem name is "smbfs". Linux filesystems are
inconsistent about taking the -fs suffix or not, but that's not
something I expect to be able to change :) (Incidentally, "autofs" is
type "autofs" because amd inserts mtab entries with "auto" in the type
field; therefore naming it "auto" would have broken interoperability
with amd. I belive it is also what Solaris uses.)
-hpa
--
<[EMAIL PROTECTED]> at work, <[EMAIL PROTECTED]> in private!
