Internal Server Error on a Heavy Loaded Server

Internal Server Error on a Heavy Loaded Server

Post by Roberto P.Martins Jr » Mon, 16 Aug 1999 04:00:00



Hi,

I'm using Apache 1.3.6 in a heavy loaded server. It serves two virtual hosts and
it's responsible for over 1 million hits per day.

But in the busiest hours I get the error page: "Internal Server Error" for some
requests page requests.

I thought Apache was limited by the MaxClients directive that was limiting it to 100
clients. Then I increased it to the maximum allowed: 256. But soon I got again
the error page... I changed the Apache source code in linux.h header, as
pointed by Apache, to increase the number os clientes served to 5000 and
recompiled Apache.

I know in the busiest hour there are 350 to 400 httpd proccess but I'm still
getting "Internal Server Error"...

I'm thinking there is some system resource limitation. I read the Apache
Performance Tips page but I didn't find any clue...

Please, does anybody knows how to help me?

Yours Faithfully,
Roberto.

 
 
 

Internal Server Error on a Heavy Loaded Server

Post by dmaho.. » Mon, 16 Aug 1999 04:00:00


Quote:> I'm thinking there is some system resource limitation. I read the
Apache
> Performance Tips page but I didn't find any clue...

> Please, does anybody knows how to help me?

What OS are you running?  I ran into similar problems running FreeBSD
2.2.8, and found the problems became a lot less severe when I recompiled
with a larger value for maxusers. Turns out I was exhausting my mbufs.

You might also be hitting your maxprocs limit - if you're executing too
many CGI's they could be using up all the available processes.

Sent via Deja.com http://www.deja.com/
Share what you know. Learn what you don't.

 
 
 

Internal Server Error on a Heavy Loaded Server

Post by Carl Byingt » Mon, 16 Aug 1999 04:00:00


-----BEGIN PGP SIGNED MESSAGE-----



Quote:

>I'm thinking there is some system resource limitation. I read the Apache
>Performance Tips page but I didn't find any clue...

Running out of kernel file handles?

- --
PGP key available from the key servers.
Key fingerprint 95 F4 D3 94 66 BA 92 4E  06 1E 95 F8 74 A8 2F A0

-----BEGIN PGP SIGNATURE-----
Version: 4.5

iQCVAgUBN7cZ7tZjPoeWO7BhAQF9LgP7BSbuu0FBAnAztsY7kKMZUalq6U/2NJmj
9PYR23ngLNb5BywMS97eiMxSkxCgCV7A6bnghx8mgzvE4VryK/ILT83kszXE+Qp9
Ut6EY5xbYOXvCpddzxQ8K6ad3FUqkgDwf9pwq0PW49nGlGN2rhMxjDptI/zJiwao
mV78iYthX9o=
=n1qu
-----END PGP SIGNATURE-----

 
 
 

Internal Server Error on a Heavy Loaded Server

Post by Roberto P.Martins Jr » Tue, 17 Aug 1999 04:00:00



>What OS are you running?  I ran into similar problems running FreeBSD
>2.2.8, and found the problems became a lot less severe when I recompiled
>with a larger value for maxusers. Turns out I was exhausting my mbufs.

I'm using RedHat 5.2 with kernel upgraded to version 2.2.5 on a dual Pentium II
450MHz with 1 GB of memory. I recompiled Apache with MaxClients modified to the
limit of 5000 clients.

Quote:>You might also be hitting your maxprocs limit - if you're executing too
>many CGI's they could be using up all the available processes.

>Sent via Deja.com http://www.deja.com/
>Share what you know. Learn what you don't.

I also execute too many CGI's. Probably in every request served there's some
CGI processing...

Thanks,
Roberto.

 
 
 

Internal Server Error on a Heavy Loaded Server

Post by Joshua Sliv » Tue, 17 Aug 1999 04:00:00



Quote:> Hi,
> I'm using Apache 1.3.6 in a heavy loaded server. It serves two virtual hosts and
> it's responsible for over 1 million hits per day.
> But in the busiest hours I get the error page: "Internal Server Error" for some
> requests page requests.

[...]
> I'm thinking there is some system resource limitation. I read the Apache
> Performance Tips page but I didn't find any clue...

The first thing to do is to look in the error log during these times
and see what the actual error is.

--
Joshua Slive

http://finance.commerce.ubc.ca/~slive/

 
 
 

1. no_root_squash made CPU in heavy load on RH 7.3 NFS server

I found that the CPU stayed in very high load when I set up a NFS
server on a RH 7.3 with the following options:

/etc/exports:

/usr/local/share           10.0.2.0/255.255.255.0 (rw,no_root_squash)

The CPU load of nfsd from `top` was always between 5-10.

While I change the option to (rw,root_squash), the CPU load decreased
dramatically to 0.1-0.5.

Actually this NFS server only serves 4 Linux clients,  so the scaring
CPU load made me
very confused.

Here are the basic information:

server ~% uname -a
Linux server 2.4.18-5 #1 Mon Jun 10 15:38:12 EDT 2002 i686 unknown

server ~% cat /etc/redhat-release
Red Hat Linux release 7.3 (Valhalla)

Any idea why?

Thanks,
Rowsen

2. Printing in Landscape

3. Apache server reliability under heavy load conditions

4. unix hard drive chg. to dos

5. no_root_squash made CPU in heavy load on RH 7.3 NFS server

6. Logitech Mouseman problems with Slackware 1.1.2

7. Our server dies when the network is heavy loaded

8. filtering users by mac address and limiting bandwidth to each mac.

9. linux server goes nuts under heavy load

10. After rebooting sendmail servers, internal servers must fully qualify for mail

11. Advice on Linux and internal web server and file server

12. How best to route internal clients to internal web server?

13. Internal Network can't browse the internal Web Server!!!!