Just before I first start running this test, the total memory in use hover
around 36,000k out of 228464. Physical Memory Available hovers around
6000k. When radlogin is started, it may jump up to 37,000k. The longer the
radlogin runs the higher the memory in use go up gradually. Physical
memory available stays steady. After about 90,000 logins, it'll go thru
15-20 (time:70) logins, then appears to get stuck on a login then cycles
thru 15-20 more. each time that occurs, the time gets longer and longer on
the login it gets stuck on. Only then, the CPU usage will drop to 11%
after running 100% all the other times.The memory in use is about 95,000k
at this point. The memory does not come back then either. The physical
memory available was stable thru all this till the slow down occurred, then
it is round between 100k - 500k the entire time.
By about this time I have to abort, cause it might take a year to let it
run thru all the the logins. Then the memory comes back.
Greg
----------
> From: Dale E. Reed Jr. <daler@iea.com>
> To: ntisp@emerald.iea.com
> Subject: Re: Radius eating up memory?
> Date: Friday, May 09, 1997 4:43 PM
>
> Greg Buffaloe wrote:
> >
> > Been testing radius by runnning 'radlogin greg greg 500000'. Below are
the
> > some of my observation. It appears to eat up memory, longer the test
run.
> > Any ideas why this might be occuring?
> >
> > Running:
> > Pentium 133
> > 32 MB ram
> > Win NT 4.0 sp2
> >
> > after about 10,000 logins. the mem has jumped up to 42,000k (no
slow down)
> > after about 18,000 logins the mem is now at 47,000k (no
slow down)
> > "" 40,000 " " " 60,000k (no
slow down)
> > "" 60,000 " " " 70,000k (no
slow down) physical mem 6400k
> > 92,000 89,000 (no
slow down) 500k
> >
> > around 93000 it stop/slows down. CPU had been running 100% now it's
> > dropping to 11%when it slow down on a single log in after about 20 fast
log
> > ins.
> >
> > still running after 98,000 logins. approx 20 logins will occur fast
> > (time=60) then on one login it will slow way down(time=100,000). It
will
> > cycle thru this from then on. The CPU usage is at 100% when the fast
> > logins occur, but drop to as low as 11% when on the slow login. The
memory
> > hovers around 100k-500k.
>
> Its caching information. I believe the memory returns a couple seconds
> after
> the radlogin stops. This is NOT RadiusNT doing this (or the memory
> wouldn't
> return), BTW. Let me know if the memory is not returning.
>
> --
> Dale E. Reed Jr. (daler@iea.com)
> _________________________________________________________________
> IEA Software, Inc. | RadiusNT, Emerald, and NT FAQs
> Internet Solutions for Today | http://www.emerald.iea.com
>
> ----------------------------------------------------------
> NTISP Mailing List listserver@emerald.iea.com
>