> Quoting Jukka Marin,
> 
> > How well do the passwd routines survive a load like this?  What is
> > the maximum UID - is it 32767 or 65535?  If 32767, it's probably too
> > low.. :-I
> 
> uid_t is 32 bits these days, so that shouldn't be a problem. Unless there's
> some old leftover of the 16 bit uid days that noone noticed before, of course

That won't be your biggest problem with that many users, actually.

Unless someone has optimized the way passwd database manipulation is
done in NetBSD, every time one of those users changes their password
or otherwise causes the database to be rebuilt, great delays will
occur as your 30K password database is rebuilt.

If memory serves me correctly, someone in the ISP world also changed
the FreeBSD system to do direct insertion of password changes rather
than an expensive rebuild-from-the-top, and perhaps these have already
long-since been merged into NetBSD.  I don't have easy access to a
NetBSD system to check, however, and that's sort of a problem when it
comes to verifying things like this.  Perhaps someone in the NetBSD
project would be willing to trade passwd entries? :-)

					Jordan