> > # Test if this user.acct session is already logged on
> > # This user.acct is not logged on, check IP address under
> > # different user
> > # Note: each session's IP address is kept under /usr/ip/xx.xx.xx.xx
> >    # build the IP addr file to capture this user's IP address
> >   build /tmp/ip/!hpremipaddr;rec=-1,,f,ascii;disc=1
> > ...
>
> The only fly-in-the-ointment I see is how does this file "go away" when
> you log off?  If you rely on a BYE udc, you could get into situations
> where a user gets ABORTed, but the file would still be left around, and
> thus the user couldn't [legitimately] log back on.

Good point! My approach works better if the logon UDC traps the user
into one or more apps but does not let the user out to the CI, etc.
This could be a situation where the NEWCI command is useful.  Short
of that, there was a suggestion of using LISTFILE ci.pub.sys,8 and
parsing out the remote IP addresses.

> Secondly, what about "aggregators" [also known as "NAT"'s] where several
> users are funneled through the same [external to them] IP address [such
> as putting a remote office on a DSL line...]

Another good point, and this time it seems the LISTFILE,8 approach
fails too... If truly different users share the same IP address
then one cannot key of off hpremipaddr. I wonder if any of the
HPSTDIN_@ or HPVT_@ variables would provide differentiation?

No good answer yet...  Jeff

* To join/leave the list, search archives, change list settings, *
* etc., please visit http://raven.utc.edu/archives/hp3000-l.html *