Centos 5.3 64 bit problem

B

BoBBYI986

Geek
Hi all, I got a problem with Centos I installed the OS on one 160gb hitahci sata drive and have two wd blue's 500gb in a software raid 1. After setting up centos it then rebooted as normal. When booting back up it starts to load the OS and then the screen just goes blank and monitor goes in standby mode. WTF?
I rebooted again same thing, then i went into bios messed with a few settings it then worked made it onto logon screen and logged in it works perfectly fine. I then got all the updates and re-booted as requested and same thing again monitor goes blank. system just completely crashes.
I did a bios update still no luck. I then installed windows xp on a spare hard drive and also ran prime 95 and memtest it passed with no problems. Xp runs like it should. So im wondering just wtf is going on? is it linux? is it the hardware im using generic driver compatibility issues?

after a few reeboots it sometimes loads the OS and it works perfect. So I've left the server on overnight to see if it crashes or anything whilst logged onto the OS.

The servers just gonna be used for domain controller and samba file store server.

Hardware:

Asrock G43 Twins -full HD
cpu: Intel core 2 quad q8200
ram: crucial 4gb

any help will be much apreciated. thanks
 
Red Hat / CentOS use quite an older kernel, since it is supposed to run on stable, primarily tier-1 hardware in a business environment. ASRock is not exactly know for their reliability or compliance to standards, so it's entirely possible they just don't play nice together. But, first things first, we should try to figure out what type of problem you're having. When the screen goes black, is the system reachable from the network (e.g. can it be pinged, SSH'd to, etc)?
 
Hi, thanks for your reply. Yea I've tried pinging the server when the screen goes blank i get nothing, also tried ssh wont log in. so it's just crashing.
Also sometimes when it crashes/screen goes blank I hit the reset switch and it restarts so the kernel is getting the signal to restart.
 
It's probably a buggy implementation of ACPI or IRQ steering in the mobo. Try interrupting GRUB and using some of the following boot options (one at a time, of course):

  • irqfixup
  • irqpoll
  • noapic
  • nopcmcia
  • apm=off
  • pci=noacpi
  • acpi=off
  • lba32

If any of these work it can be the basis for further troubleshooting.

Out of curiosity, why CentOS? Is it for a server lab exercise, or are you trying to use it for a desktop OS?
 
I will give it a go tomorrow when I get back to work and I will post back with results. Thanks alot for your help, much apreciated.
 
Hi it's just what I get instructed to use from the boss. It's also free to use linux other than using windows server. So overall this reduces the cost for our clients. All it's gonna do is be a domain controller and filestore.
I know it's not proper server hardware but the money the clients want to spend is nowhere near enough to buy the hardware for me to build them a Xeon server.
 
I see. CentOS is a good choice then, though you will have to tweak Samba out of the box a bit since their default rollout is a bit broken. Still, after a little TLC it works great, and I've found it outperforms a "real" Windows server about 2.5:1.

Also, I have a lot of experience consulting for small shops with a shoestring budget. If you won't find me too arrogant for saying so, here's a few tips for building budget servers in the future:
  • Use motherboards that are as close to the reference design as possible to assure compatibility and stability. In other words, models with names like "1337 Overclocker Blade Dragon 9000!!1" are not a good choice. Stuff by SuperMicro or Tyan, on the other hand, are an excellent choice.
  • Try to get components that use solid capacitors instead of the traditional liquid caps. They will last a lot longer; nobody likes to replace a server unexpectedly. This is especially true of motherboards, but applies to other things like NICs and RAID cards as well.
  • Do a little research on the chipset against the operating system of your choice for potential problems before you buy. Stability is king, and untrustworthy components are your enemy.
  • Most budget servers you build from off-the-shelf parts aren't going to have anything so extravagent as a redundant PSU, but at least overspec the consumer-grade PSU more than you normally would with a workstation. Overspecing will help the system to run for longer more reliably, and you never know when you're going to have to add more HDDs to a server.
  • A fancy 3ware or LSI SATA RAID may not be an option, but avoid using "fake-RAID" driver-based RAID solutions, e.g. NVidia or Adaptec. These are much more unreliable than an OS-based software RAID and should be avoided like the plague.
  • Good cooling is very important for long-term stability. If you're using a chassis that requires lots of fans, make sure they're ball-bearing and that all intakes have filters. It doesn't cost that much more to get a half-decent chassis, and it makes a big difference in stability, longevity, maintainability, and presentation.
  • On the topic of presentation, make sure that all visible drivebays match in color and style with the chassis. Avoid "gamer" chassis with windows or funky plastic faceplates; these will cast a shadow over your professionalism and technical competencies.

P.S. - These articles may help you: anti-trend.homelinux.org - Linux Security Articles
 
Hi thanks for your help and tips. I checked the server this morning it was still operating. So I restarted it a few times also turned it off for a good 10 mins it booted up everytime no problem. I noticed something last night all the hard disk l.e.d's where all a constant red whilst logged into centos. So I presume the OS must of been rebuilding the raid? But when i come to it this morning no constant red hd l.e.d's, everything was fine.
 
Interesting, but it still should have been able to boot fine with the RAID rebuilding. You can check the status of a Linux RAID like this:

Code: 
cat /proc/mdstat
 
Thats what I was thinking it should rebuild the software raid fine because the OS is on a seperate disk. I will run that command when I get back to work next week. I will be copying the data across next week to the server, also have to install a pci scsi controller for a external tapedrive for backup.
 
You must log in or register to reply here.
Back
Top