Re: sendmail erro

From: Alex Rousskov (rousskov@measurement-factory.com)
Date: Sat Mar 03 2001 - 11:02:50 MST


On Sat, 3 Mar 2001 chin60@pchome.com.tw wrote:

> When FreeBSD sends mail to server or client.

I ma not sure what you are saying here. Polygraph has nothing to do
with your e-mail setup. Polygraph agents cannot send or receive
e-mail.

> Because the IP addresses are virtual, these IP address don't have
> domain name.

IP aliases may have domain names if you configure your DNS to resolve
them. However, I do not understand why any process in you system would
want to send mail to Polygraph-specific IP aliases.

> So it always has sendmail error on system. Would this
> error influence our polyclt and polysrv? How do we correct this
> error? Shutdown mail demaon? Or there is option to choose in
> polysrv and polyclt?

You can shutdown the daemon or reconfigure the program the emits
e-mails (whatever that is). Polygraph has nothing to do with e-mail.

> Because our squid always can't pass through your benchmark. It
> always be interruped by polyclt's auto- shutdown. And polyclt
> still has synchronal problem with polysrv. Does cache disk size,
> cache disk memory and main memory influence that? Could you give
> me some advice? Thanks1

IIRC, your experiments showed poor (20+seconds) response time causing
Polygraph client to run out of memory. This is not a problem with
Squid or Polygraph -- you are just overloading your Squid. That is,
you tell Polygraph to create loads that your Squid installation cannot
handle. You may want to decrease the load to, say, 1 req/sec and see
if it solves the problem. If it does, then you can gradually increase
peak load to find the maximum that your Squid installation can
support.

And yes, cache and memory sizes affect Squid performance. You can read
cache-off results to see what some of the Squid configurations can do
and compare those configurations with yours.

Alex.



This archive was generated by hypermail 2b29 : Tue Jul 10 2001 - 12:00:18 MDT