On Sun, Feb 13, 2005 at 07:30:35PM +0300, Denis Lagno wrote:
> > It is the same reason. The umount command tries to access the
> > old mountpoint and fails because that mountpoint is invalid.
> > 
> > You can use the 'umount -R' command to bypass "all attempts to
> > be smart about mechanically determining the correct path from
> > the argument".
> 
> *sigh* It also does not work in this case:
> 
> # mount | grep stuffy
> flam:/stuffy on /flam/stuffy type nfs (read-only, nosuid, nodev)
> # umount -R flam:/stuffy  
> umount: flam:/stuffy: No such file or directory
> # umount -fRv flam:/stuffy
> flam:/stuffy: unmount from flam:/stuffy
> umount: flam:/stuffy: No such file or directory
> # umount -fRv /flam/stuffy
> /flam/stuffy: unmount from /flam/stuffy
> umount: /flam/stuffy: Invalid argument
> 
> 

I recall that you need to do something to the RPCdaemonic stuff.. I am every-
thing else than a NFS-user, do the clients even have a RPCd? Killing that helps?
Or the server-side.. . I don't even really know the idea behind it, I could
just guess.