Koozali.org: home of the SME Server

System stability.

Zee Naushad

System stability.
« on: April 03, 2002, 08:34:16 AM »
I'm extremely happy with e-smith in general, I've managed to tweak it and twist it in a way that I like, mainly by installing several new rpms and such.  However, lately it's been giving me a lot of headache.

Mainly from the fact that it crashes every 4-5 days or so.  At one point my uptime was close to 180 days, but now I am lucky if I see a week.  

Now before anyone says it's hardware related, I assure you 100% it isn't.  I've swapped components several times, new nic(s), new videocard(s),new motherboard(s),new cpu(s), and even new memory.  

I run a website that's fairly well visited off my server.  I get about 3000 hits a day.  I also run a distributed.net proxy program.  My site relies heavily on php and mysql.  http://www.hardwareflux.com

I need to find out why e-smith is crashing on me so much.  

Can anyone suggest where to even start?  

I'll explain the crash.......I have my terminal set up on TTY2 where I log in as a user and run the distributed.net proxy program.  I issue a setterm -blank 0 command so the screen doesn't blank, and that's about it, I let the screen just stay there and watch the proxy output from time to time, while I work on my main machine.  It works flawlessly, and I do mean really well, for about 4-5 days, and then boom, just freezes, without any error message or what not.  Just a solid lock.  Have to do a hard reboot for the system to reboot.  It does through the horrid fsck process and takes a good 5 minutes to boot.

Same thing happens after 4-5 days again..

Plz help me, I'm going nuts!

Des Dougan

Re: System stability.
« Reply #1 on: April 03, 2002, 09:58:59 AM »
If you're convinced it's not hardware, you may want to check your httpd logs. Our production web server crashed on Saturday. The last (corrupted) entry in the apache logs was from the Scooter web spider. I found the link below today - note the comment about memory overuse:

http://www.webmasterworld.com/forum1/1394.htm


Des Dougan

Filippo Carletti

Re: System stability.
« Reply #2 on: April 04, 2002, 05:08:45 PM »
To get an idea on the source of the problem try using Magic SysRq.

If you don't know what magic sysrq is, begin enabling it:
1. edit /etc/sysctl.conf
# Disables the magic-sysrq key
kernel.sysrq = 1
2. sysctl -e -p /etc/sysctl.conf

Verify it works.

When your server seems to have crashed, use magic sysrq from the console.

Info here:
http://www.linuxhq.com/kernel/v2.4/doc/sysrq.txt.html

Zee Naushad

Re: System stability.
« Reply #3 on: April 09, 2002, 05:01:30 PM »
Thank you for the tip regarding sysrq, I will try to check it out when and if the server crashes!

Zee Naushad

Re: System stability.
« Reply #4 on: April 12, 2002, 01:22:34 PM »
I hate to say this, but the culprit turned out to be a faulty PCI videocard.

Grumble.......

Sorry about all the hooplah.