In some mail from Mario Peschel, sie said:
[...]
> First of all I think you must define what an inactive user is.
> It seems to me that the best definition of an inactive user is
> that data is neither coming from or going to the users terminal.
> This means that the program (or whatever) should test both a terminal's
> access and modification time before logging a user off.
> The last interaction with the terminal would be the most
> recent of the two times.
This is a totally bogus check.
I've got a program around somewhere called "ni". When I ran it, users would
speculate as to what it was for I was never idle. That is what it is an
acronym for.
If a user doesn't want to be seen as being idle then you're probably going to
have to use snoop/tcpdump/ttywatcher to work out if they are or aren't.
The things you learn when you're a student and want to avoid dialin idle
timeouts O:-)
darren
Follow-Ups:
References:
-
Re: Watchdog
From: "Mario Peschel" <Mario .
Peschel @
mch .
sni .
de>
|
|
