Subject: | |
From: | |
Reply To: | |
Date: | Thu, 31 Oct 2002 12:58:58 -0800 |
Content-Type: | text/plain |
Parts/Attachments: |
|
|
On Thursday 31 October 2002 11:21, Michael Berkowitz wrote:
> Tim Atwood writes
>> Anyone have a brilliant idea how to get a high-water mark of the number of
>> sessions on at any one time? Say during a one month period?
[...]
> ---------------------------------------------------
> As part of our user logon security system (home grown), we capture the
> value of the variable HPSESCOUNT at session logon time. This shows the
> number of users currently logged on. Simply capture this into the data
> base of your choice.
Along those lines, you could set up a job that runs "all the time" with a
simple while-forever loop:
setvar hiwater HPSESSCOUNT
while true
setvar x hpsesscount
if (x > hiwater)
setvar hiwater x
endif
pause <some interval>
endwhile
capturing "hpsesscount" into the variable "x" avoids the situation where
sessions drop off between "if hpsesscount > hiwater" and "setvar hiwater
hpsesscount", thus giving you a false hiwater reading. HOWEVER, due to the
fact you don't have infinitely small granularity because of the pause, this
is kind of a moot point :)
Then again, the other problem you have is in getting this "hiwatermark"
information out of the job... ("while true" takes soo long to execute -- I
guess you'd have to get an uncrippled HP-UX processor to finish it in some
reasonable amount of time ... :)
But, the "wow, I coulda had a V-8" solution is to scan your log file(s) --
every time you get a "session start" record (102? 103? can't remember), bump
up the counter and check against your "hiwater mark"; every time a session
termination record comes up you decrement the variable. [system
crash/restart records obviously reset the "current count" to zero...] Of
course, if your current count goes negative, start checking for cracker
activity...
* To join/leave the list, search archives, change list settings, *
* etc., please visit http://raven.utc.edu/archives/hp3000-l.html *
|
|
|