RID <-> UID map file...why?

[Ray Strode]

Hi everyone,

Why does winbindd use a tdb file that maps rid's to dynamically 
generated uid's?
I ask because with the 2.4 linux kernel uid's are 32-bit values, so it 
seems that
the rid could be used for the uid and gid directly.  Is this because 
rid's might
conflict with system uid's (< 500) ?

I'm asking because I may be setting up a lab of linux machines, that 
would ideally
be able to authenticate against a Windows NT PDC and they will be using an
nfs mounted /home.  This means that the uids of the users need to be 
consistent
accross machines.  

If i mount a common locks directory on all the machines, will all the 
different
winbind daemons coexist together?  Or will I end up with a corrupted map 
file?

Are there any good ways to do what I want to do, or is winbind really 
only designed
for one machine?

(Please CC me, I'm not subscribed)

--Ray Strode