[mariolima]

Hi:

I will detailed describe a serious problem we are facing with one of our servers, and I thank a lot in advance if someone who already

has experienced such a problem could help us.

Summarizing, we had an email Server running under LINUX, which we replaced by a better machine. Instead of having better

performance and better service levels, we have some error messages besides seeing higher load average times and slowing down.

I will now describe the former and the new environments as well as the real problem:

FORMER:

Processor: Pentium IV 1.8 Ghz
Main Board: Asus P4V533
Memory: 1 GB DDR
HD: Seagate IDE 40 GB
Network: Encore

Operating System: Linux Red Hat 8
Some additional software: Apache, Qpopper, Sendmail, Pop-before-Smtp, SpamAssassin

NEW:

Processor: Pentium IV 3.0 Ghz HT FSB 800
MainBoard: Asus P4V800
Memory: 1 GB DDR PC3200 (400 mhz)
HD: Seagate Serial ATA 120 GB
Network: Intel Pila 8460

Operating System: Linux Fedora Core2, Kernel 2.6.6-1.435smp
Some additional software: Apache, Qpopper, Sendmail, Relayer, SpamAssassin

PROBLEM:

On the former environment we had, of course, moments of great peak, where those Load Average Times (commands uptime or top)

went to incredible 30.x and higher. But, normally, although it had great activity for it was an emails server, it worked with good

service levels.

But now, with a really more powerfull machine and a brand new operating system, we are facing those terrible times practically all

the day long. And we feel that our daily common operations like reading/sending emails, SSH, Ftp, ordinary Linux commands are

sensibly delaying with poor response times. Besides this, dmesg command is showing lots of occurrences of error messages like

this:

hdg: dma_timer_expiry: dma status == 0x24
hdg: DMA interrupt recovery
hdg: lost interrupt

So, please, if someone could see on this report details a reason or what the problem could be, please, help us as soon as possible,

for we are in real troubles, because this is a very critical server for us.

Thanks.

Mario Lima./

[Squashman]

Originally Posted by mariolima

So, please, if someone could see on this report details a reason or what the problem could be, please, help us as soon as possible,

for we are in real troubles, because this is a very critical server for us.

Thanks.

Mario Lima./

It looks like a kernel problem to me. From what I have read on Red hats site, Kernel 2.6 doesn't like to play nice with SATA or the motherbaord might be bad or needs a bios upgrade.
http://www.redhat.com/archives/fedor.../msg00010.html

Do you realize that Red Hat throws all their experimental stuff into Fedora to test it before it ever moves to Enterprise Server. Why would you want to use Fedora for a Production server. That is just suicide.

[mariolima]

I'll be searching this. Please, if someone else has something to add, I thank a lot.

Mario./