I tried... but again I'm dealing with a virtualized environment and penguin 1 cant see if penguin 2 is using 50% of the CPU. Thats all handled by the VM hipervisor. Never had enough individual system load to make nice show me any difference. Figured nice was a long shot at best but I do not know of any other methodology to put the brakes on something within a specific linux instance.
Trog
<[EMAIL PROTECTED]>
Sent by: To
clamav-users-boun ClamAV users ML
[EMAIL PROTECTED] <[email protected]>
net cc
Subject
01/26/2005 01:48 Re: [Clamav-users] Using Clam AV -
PM Perhaps I am not understanding
product intent
Please respond to
ClamAV users ML
<[EMAIL PROTECTED]
ts.clamav.net>
On Wed, 2005-01-26 at 13:17 -0600, [EMAIL PROTECTED] wrote:
> The first part I am not sure how to do. The second is easy enough,
however,
> when I used clamdscan the file system scan consumes inordinate amount of
> CPU resources. I've tried starting clamd with a nice value of 17 and
> running clamdscan with a nice value of 18, in hopes of slowing it down so
> that the consumed CPU cycles are spread over more 'real time'. Why is
this
> important? In a virtualized environment such as the one I am running in,
> memory and CPU resources are currently being shared by about a dozen
> virtual servers. If one single server consumes 40-80% of cpu resources (2
> processor configuration), the act of running the scan on all systems is
> going to completely bury the box.
>
I haven't tested nice'ing clamd, but how did you test it? Nice'ing won't
make a noticeable difference until there is contention for the CPU, did
you try some benchmarking while clamd as running nice'ed against
normally?
-trog
(See attached file: signature.asc)
_______________________________________________
http://lists.clamav.net/cgi-bin/mailman/listinfo/clamav-users
signature.asc
Description: Binary data
_______________________________________________ http://lists.clamav.net/cgi-bin/mailman/listinfo/clamav-users
