Hello,
Peter Bowyer Wrote:
---------------------
====
> host_lookup = 0.0.0.0/0
Umm, doesn't that mean you're doing a host lookup on everything? That has to
have a performance impact, even if it's looking the same hostname up every
time.
====
According to the Exim documentation, that should only run a lookup on
0.0.0.0 IP addresses which of course since there is no such IP, it won't.
That is how I interpreted.
====
> system_filter=/etc/blank
So must this - why not take the directive out altogether? There has to be
some processing going on for every mail, even if the file is empty.
Have you tested how fast your mailing list software can generate mails into
/dev/null ?
====
I did not think of completely remove the system filter, I will try. For the
/dev/null, I cannot figure out how. I tried setting the exim path to
/dev/null, but that did not seem to work because now it reports NO emails
being sent.
====
> I read on the internet that a server was able to get 42 emails per
> second with a 1.13MHz server.
Where did you read that? The internet is a big place. If you quote your
source, maybe investigations can continue.
====
Here is one of them, but not the exact one I saw:
http://groups.google.com/groups?q=exim+performance+queue_only&hl=en&lr=&ie=U
TF-8&oe=UTF-8&safe=active&output=search&selm=slrnau9iil.119.jc%40yard.minjus
t.gov.ua&rnum=2
====
> 10MPS port/switch
> 1000gb Bandwidth
Is that 1000 Gigabytes/sec or 1000 Gigibits/sec? Makes a big difference,
especially when you've only got a 10Mbit/sec switch port. ;-)
====
That is a total of 1000gb allowed for my server.
---------------------
Quoted by Russel King:
====
You do realize that you're at the mercy of the time to perform DNS
lookups yourself, the time for the remote machine to perform DNS
lookups of your IP address, the time for the remote machine to
respond to your SMTP commands etc.
====
I understand that, however I know there must be away to increase this speed.
====
However, I'm amazed that you think you need to kill off all existing
exim processes and restart them later - that sounds like a good
recipe for disaster, as well as being grossly inefficient and
detrimental to any in-progress outgoing or incoming SMTP
transactions.
====
I do this because when my script runs again, exim is trying to deliver while
it is queuing, which causes the load to go very high. When just running the
queue by it's self with up to 75 processes, I get a total of load sometimes
to 21.
-------------------------
Quoted by *Many*
====
**Using killall/pkill compared to killer**
====
I use killer because my version of killall I use does not support regex,
which I require to kill only.
---------------
Since I am using a simple IDE 7200RPM 80gb Harddrive, would it increase
performance by a good enough jump to go with a SCSI?
Thanks and God Bless!
David Garza
-----Original Message-----
From: exim-users-admin@??? [
mailto:exim-users-admin@exim.org] On Behalf
Of Peter Bowyer
Sent: Friday, May 07, 2004 7:46 AM
To: exim-users@???
Subject: Re: [Exim] Exim is Unreasonably slow
David Garza <dgtech@???> wrote:
> host_lookup = 0.0.0.0/0
Umm, doesn't that mean you're doing a host lookup on everything? That has to
have a performance impact, even if it's looking the same hostname up every
time.
> system_filter=/etc/blank
So must this - why not take the directive out altogether? There has to be
some processing going on for every mail, even if the file is empty.
Have you tested how fast your mailing list software can generate mails into
/dev/null ?
> I read on the internet that a server was able to get 42 emails per
> second with a 1.13MHz server.
Where did you read that? The internet is a big place. If you quote your
source, maybe investigations can continue.
> 10MPS port/switch
> 1000gb Bandwidth
Is that 1000 Gigabytes/sec or 1000 Gigibits/sec? Makes a big difference,
especially when you've only got a 10Mbit/sec switch port. ;-)
Peter
--
## List details at
http://www.exim.org/mailman/listinfo/exim-users Exim
details at
http://www.exim.org/ ##
