Hardware issue [solved]

[dude67]

I've been having a great deal of problems with this PC at home. This is only some 18 months old HW and it 'should' be OK and working fine but still...

 

For several months already I've had some problems booting my Mandriva 2008.0 and 2008.1. The system would start loading, but would freeze at some point - I couldn't really tell what it is that's causing it. I always thought it was some software related problem and didn't really pay that much attention to it. The main reason beeing that I would always get it up and running even though in the worst cases it would require more than five reboots to get it running. And the WinXP home I have installed in it also seems to boot fine all the time.

 

Then two weeks ago (or there abouts) it lost my /home partition! I did reinstall Mandriva 2008.1, but on the second boot it had lost the root-partition (/-partition). I was certain it was the HDD (500GB Seagate Barracuda 7200.10) that was acting up and getting ready to go to hard disk heaven. I then bought two 750GB Samsungs and made a RAID1 array out of them. Now it seems that I had it covered.

 

But now it's doing weird stuff again. It takes even more times rebooting (sometimes ten isn't enough) to get Mandy up&running than it did before. It usually ends somewhere where it says it has loaded eth0 [OK] and ath0 [OK] and then freezes. Then it may produce a very long list of some checks (with some different PIDs all the time), but I really couldn't understand what it was doing... Then in the end it would give kernel panic note - or sometimes it said that some problems were solved, but the system needed to be rebooted. But I can get Mandriva running in the end (KDE desktop - if that makes any difference) and everything seems to be operating just perfect. It has never crashed (knock, knock) when Mandriva KDE is fully loaded.

 

I have also installed WinXP home on the sda (first before the mandriva installation), but now I cannot boot there either. It seems the windows partition has also been damaged... I did the RAID1 array with Mandriva installation DVD, so I couldn't make it take the windows partition into the RAID array.

 

Any idea or suggestions what it could be? I'm suspecting the motherboard is damaged, but is it? I wouldn't want to get a new MB just to find out that the first one is working fine and the problem is somewhere else...

 

Oh, and the system logs don't reveal anything. I tried finding the kernel panic note or anything else from the failed boots, but there is nothing in the syslogs about that! I read them from MCC system logs (messages and syslog).

 

This is the hw I have:

Motherboard

MSI 945P NEO3-F

Processor

Intel Core 2 Duo / E6400 2.13GHz/2MB

HDD

OLD: 1 x 500GB Seagate Barracuda 7200.10 SATA

NOW: 2 x 750GB Samsung SATA2 (HD753LJ)

Memory

2 x Buffalo 1GB DDRII 533 MHz

Edited September 9, 2008 by dude67

[ianw1974]

First, I"d check if you're getting any memory errors by using memtest. If you are, then if your BIOS has the option choose "Load optimised defaults", and see if it performs any better after that. If not, check memtest again to see if you're getting memory errors. If more than one memory module installed, remove one, and check/test for memory errors again. If so, put the other module back in removing the other to check if the other is faulty. If both are causing errors, then either we have two faulty modules, or the mother board is faulty.

 

Since you seem to also be saying that it's crashing a lot, also check the cables to your hard disks and perhaps replace these where necessary. Maybe they are old and no good and the cause of the problems. Reseat any PCI cards as well as the processor and memory and see if it's any better.

 

Also, failing that, remove all peripherals you don't need, so any pci cards you don't need, any CD-ROM's you don't need, or anything else connected to the machine that you don't need, and test a minimal setup.

 

I also had problems with dmraid hanging my system on startup. Technically it wasn't hanging it, just pressing ESC was enough to bypass and boot the system. No way to disable it though unfortunately. I did check. But this prob isn't your problem.

[dude67]

Thanks Ian, I'll do these checks tomorrow - starting with the memtests. I believe you are referring to this https://mandrivausers.org/index.php?s=&...st&p=496544

 

The SATA cables are new (except for the one used with the old 500GB HDD that's still hooked on). But I will check them again if the memory modules (2) are working fine.

[ianw1974]

You can do that yes, or download a Gentoo Minimal X86 CD which is about 150MB and this has memtest included on it. I don't know about the Ultimate Boot Disk. You can also urpmi memtest in Mandriva, and then boot it from the grub menu to test your machine. Of course, if you can get Mandy booted first :)

[dude67]

You can also urpmi memtest in Mandriva, and then boot it from the grub menu to test your machine. Of course, if you can get Mandy booted first :)

Does it work with lilo also? I was told that grub doesn't allow booting from md0 (in my RAID1 array), so I installed and am using lilo.

[ianw1974]

Grub does allow booting from md0, just Mandriva is weird in that respect! Yes, memtest will be added to lilo, this is how I did it once with Mandrake 10.1. So should work the same now too.

[Lexicon]

1 x systems installed with Mandriva 2008.1 Spring (1 @ Home)

 

Just at last my congratulations ...Lex

Edited September 7, 2008 by Lexicon

[daniewicz]

dude67:

 

The Ultimate Boot CD is a good thing to have around for hardware diagnostics.

http://www.ultimatebootcd.com/

[dude67]

Grub does allow booting from md0, just Mandriva is weird in that respect! Yes, memtest will be added to lilo, this is how I did it once with Mandrake 10.1. So should work the same now too.

Can I use grub with this RAID1 installation as well? I quite like grub over lilo. How can I change that? Mandriva automatically changed the boot loader from grub to lilo when I made the RAID1 array during installation.

[dude67]

dude67:

 

The Ultimate Boot CD is a good thing to have around for hardware diagnostics.

http://www.ultimatebootcd.com/

 

Thanks daniewicz, I already downloaded it and am going to burn it tonight.

[ianw1974]

Can I use grub with this RAID1 installation as well? I quite like grub over lilo. How can I change that? Mandriva automatically changed the boot loader from grub to lilo when I made the RAID1 array during installation.

 

When I tried setting up raid with Mandriva, nope, it only allowed me to use lilo and nothing else. I don't know why, but Mandy disable the selection of grub. I couldn't even install it manually, because I wanted to do this also as it wasn't available in mcc. I never asked Mandriva why they did this. Every other distro doesn't have a problem with it, so why Mandy does I have no idea.

[So_it_goes]

Had similar problems on my sons computer an old motherboard.

In our case it was the battery.

 

Probably not your problem as you have a rather fresh mobo.

But anyway as a last resort. :-)

 

Started with kernel panics on memory 0x.......

Then the winXP login screen took longer and longer time

to start( no linux on this machine ).

Then it booted the bios screen instead of the Winlogin screen.

 

Changed battery and everything works OK now.

 

//

[Greg2]

Here's the section in the Errata about lilo, grub and RAID: 2008.1_Errata

 

You can also just d/l the memtest86 iso, it's only 1.8 MB. memtest86.com

[mindwave]

The SATA cables are new (except for the one used with the old 500GB HDD that's still hooked on). But I will check them again if the memory modules (2) are working fine.

 

if you had issues w/ the old 500, replaced it w/ 2 750's yet are still having problems, maybe if you disconnesct the old 500?

 

j

[dude67]

Thank you all for your suggestions.

 

@mindwave: that did cross my mind and I did try booting the PC with that old HDD disconnected (disconnected the SATA cable). No help.

 

I tried the memtest (Mandriva repos and the one from the ultimate boot cd) and the results were 0 errors. I guess that's good news.

 

Then I took out the Planet WL8310 PCI card (wlan card) as I also now have ethernet cable connected to the pc and voilá: It booted fine! I tried four times in a row and it seems to boot up just fine. I'll just wait a few days before marking this thread solved.

 

So thank you all, especially ian!

 

Again the virtuals are on me

Edited September 9, 2008 by dude67