This is not pure polygraph question, but I need your experience to help me
out here.
I'm installing FreeBSD 3.4 on Dell OptiPlex GX110 desktop machines. All
setup procedure is good but one problem - the system doesn't recognize more
then 65MB of RAM!
BIOS and NT OK with the 256MB./ FreeBSD is not.
I have installed the system on different hardware (IBM, Gateway) and they
recognize the real memory amount (128MB and 256MB).
Can anyone help me with this?
Eli Sagie
Performance Lab Manager
Cell: 054 998041
Office: 972 4 8561068
Lab: 972 4 8561067
mailto:elis@microsoft.com
-----Original Message-----
From: Alex Rousskov [mailto:rousskov@measurement-factory.com]
Sent: Friday, August 18, 2000 6:19 PM
To: Vincent Y H MAn
Cc: polygraph@ircache.net
Subject: Re: Polygraph machine slows to halt
Vincent,
You are probably running out of memory on Polygraph machines.
When proxies cannot respond fast enough, PolyMix-2 robots will queue
waiting transactions and that consumes memory. You may want to search
for "virtual memory exceeded in new" postings on this list.
As a rule of thumb, you must ensure that polyclt/srv processes
do not page. Another important thing is to monitor memory allocated for
kernel network buffers. If that memory is exhausted a machine can become
unusable (try "netstat -m" on FreeBSD).
Alex.
On Fri, 18 Aug 2000, Vincent Y H MAn wrote:
> I've been using polygraph 2.2.9 (polymix-2 workload) trying to
> benchmark some caching servers. As I was testing, the caching
> server couldn't allocate enough memory, and shut down the ethernet
> connection. After I ifdown/up it started working again, but the
> polygraph machine would slowly creep to a halt. It didn't crash, it
> just appears like the machine is frozen. But if you ifdown the
> (caching) server again, the polygraph machine would be back to
> normal (until you ifup the server). I really don't have any other
> information to give, as the only 'errors' were on the caching
> server, and not on the polygraph machines. Any ideas?
>
> Thanks!
>
> Vincent
>
This archive was generated by hypermail 2b29 : Tue Jul 10 2001 - 12:00:14 MDT