Thanks to :-
William Beelen <william.beelen_at_digital.com>
N V Brahmaji Rao <nvbr_at_iitk.ernet.in>
Alan <alan_at_nabeth.cxo.dec.com>
Bob Jones <BJ_at_oregon.uoregon.edu>
Tom Webster <webster_at_ssdpdc.lgb.cal.boeing.com>
Caine <ctflist_at_jane.penn.com>
Jerry Berkman <jerry_at_uclink.berkeley.edu>
Original Question :-
> We have an account creation ksh script here that is a big time saver.
> I've expanded it's functionality alot from some code that someone
> left behind. One thing that I have left in from the older code is
> a line that limits the UID number of the created account to < 8000.
> Only because I don't understand why it is there.
>
> The script eventually generates a passwd NIS map.
>
> We are not using passwd shadowing.
>
> We are now fast approaching uid 8000. Can anyone think of any reason
> why I shouldn't remove this limitation ?
Summary :-
The subject of my mail was misleading. I wasn't really after the figure
for the ultimate max uid. I was more after a reason why someone would
limit the number of uids on a system to 8000. Anyway...
The general consensus was that 8000 was way below any limit I might
want to set.
Responses :-
Tom Webster:
"Unless there is a site dependant reason, you should be able to set the
limit up to the 60000 range (60001 and 65534 are used by the nobody users)."
William Beelen:
Quoted "Release Notes for DUNIX v4.0D. Page 1-3."
"With Digital UNIX Version 4.0D, the maximum number of UIDs and
GIDS allowed increases from 65,535 (16 bits) to 2,147,483,647 (31 bits).
Note that this increase does not mean that 2.14 million users can
simultameously log onto a system... ...The maximum number of users that
can be logged on is determined by the available system resources."
Brahmaji:
"The response time on NIS look up etc will be high if the map is very
big. To my knowledge, no other effects."
Alan:
"Older versions of Digital UNIX use a short as the data size
for a UID. So a practical limit is 32768 or 32767. I think
V4.0D allows using 32 bit integers, but some applications
may still be compiled to use a short as the size."
================================================================================
Thanks again,
Richard Price.
(richard.price_at_vf.vodafone.co.uk)
Received on Fri Sep 11 1998 - 09:32:45 NZST