Is there any database involved?

This would be a likely culprit.

I cannot see it being apache past others recommendations of changing to worker.

The keepalive thing concerns me as well.

Excuse the out of form response.



----- Original Message -----
From: "Krist van Besien" [EMAIL PROTECTED]
Sent: 04/01/2008 04:26 AM
To: users@httpd.apache.org
Subject: Re: [EMAIL PROTECTED] need 'high traffic' config



On Tue, Apr 1, 2008 at 5:07 AM, James Mandy <[EMAIL PROTECTED]> wrote:

>  The server has plenty of cpu, 4Gig ram, RAID, etc. It's fast. During
>  busy periods, requests are taking a long time. checking the server I
>  see it's not under much load and there's usually always still 2Gig of
>  memory free on the machine. I suspect that clients are 'waiting' for
>  access to apache when the machine is getting a lot of hits. I'm
>  thinking my configuration is just wrong.

How many apache processes do you see when the server is under load?


>  I'm wondering if there's anyone here who runs a high-traffic server
>  like ours who might have some experience in tweaking apache for this
>  kind of work? I'd love to chat with you. We are even willing to pay
>  for someone to help us dial this baby in if need be.
>
>  I'm happy to discuss the finer details of our infrastructure with
>  someone that would be willing to help us. Here's a little look at some
>  of our httpd.conf:
>
>  Timeout 20
>
>  KeepAlive Off  (turning this on seems to make things crawl)
>
>  <IfModule prefork.c>
>  StartServers 8
>  MinSpareServers 15
>  MaxSpareServers 20
>  ServerLimit      500
>  MaxClients       500
>  MaxRequestsPerChild 10000
>  </IfModule>

If you have lots of memory to spare I'd increas MinSpareServers. That
way more processes are ready to receive requests. If you have an
apache with lots of modules loaded (php?) the startup time of a new
server process is not trivial. So better make sure that you have
enough spare servers to handle bursts in traffic.

Also you might want to increase both ServerLimit and MaxClients. You
might want to investigate using the multithreaded "worker" mpm.

>  Also getting a lot of this in /var/log/messages: (all the time)
>
>  Mar 29 11:05:57 sls-hc16p2 kernel: possible SYN flooding on port 80.
>  Sending cookies.
>  Mar 29 11:05:57 sls-hc16p2 kernel: TCPv6: dropping request, synflood
>  is possible

You might want to add:
ListenBackLog 128
to your webserver config.

Krist



--
[EMAIL PROTECTED]
[EMAIL PROTECTED]
Bremgarten b. Bern, Switzerland
--
A: It reverses the normal flow of conversation.
Q: What's wrong with top-posting?
A: Top-posting.
Q: What's the biggest scourge on plain text email discussions?

---------------------------------------------------------------------
The official User-To-User support forum of the Apache HTTP Server Project.
See <URL:http://httpd.apache.org/userslist.html> for more info.
To unsubscribe, e-mail: [EMAIL PROTECTED]
   "   from the digest: [EMAIL PROTECTED]
For additional commands, e-mail: [EMAIL PROTECTED]


---------------------------------------------------------------------
The official User-To-User support forum of the Apache HTTP Server Project.
See <URL:http://httpd.apache.org/userslist.html> for more info.
To unsubscribe, e-mail: [EMAIL PROTECTED]
   "   from the digest: [EMAIL PROTECTED]
For additional commands, e-mail: [EMAIL PROTECTED]

Reply via email to