MDK10.1 Lockup During Runlevel 3
 

I'm at witts end so I thought I would pose the question here to get some ideas.

I'm running Mandrake 32bit 10.1 on an AMD64 CPU. Stock Kernel (2.6) and a resonably recent urpmi update.

On the server I am running an IDE drive for the OS and 2 SATA drives in RAID 0 for the data.

The kernel presently loads the following kernel modules:
raw 6528 1
md5 3584 1
ipv6 230916 47
af_packet 16072 2
floppy 55088 0
r8169 14180 0
ide-cd 37280 0
cdrom 37724 1 ide-cd
loop 12520 0
supermount 34804 1
sd_mod 19232 2
sata_via 4836 1
libata 38308 1 sata_via
scsi_mod 104044 2 sd_mod,libata
genrtc 7444 0
ext3 120680 2
jbd 49080 1 ext3

Services running are: (Off the top of my head).
Courier/IMAP
Postfix
MySQL
Apache 2 /w SSL running Horde
Samba

As you can see, nothing special is loaded. Very stock stuff. No X, no NFS , no xfs. Nothing.. Just bare bone network services and kernel modules.
Iptables is not installed and running. So it's not a firewall issue

The problem I am encountering is a freeze. The console will lock up and accept no entry, the network services will stop answering *but* I am still able to ping the system and get a reply back. I've checked dmesg / messages / secure and there is no mention of ANYTHING regarding a problem. The only noise I hear is Samba complaining about CUPS not being installed. The only way to resolve the problem is to coldstart the system (which sucks as it means I have to drive into work at 3am in the morning).

The only thing I can think of is a problem with the kernel. Some kind of driver which is interacting with the hardware and causing it to lock up. Would this kind of error cause the issues I am encountering? ie: Locked Console / Dead services but ping replies? Are there any problems with certain motherboards and MDK 10.1? I checked the internet before ordering the system and I didnt' see any red flags. Assuming this thread has some interest, I can delve deeper and get all the Motherboard / hardware information for further review.

I hope someone can help as the only recourse is to dump Linux and switch to a Windows Exchange setup which I am trying to avoid like the plague.

thanks,

-- Me

Try booting with
(you can also put these options in your lilo.conf)

Also have you got all the latest updates using urpmi?

I wouldn't say this was your only recourse. If you can't solve the stability problems with Mandrake then you should try one of the Redhat Enterprise Linux clones such as Centos, or even RHEL itself. Although Mandrake is my favourite distro I'd still put money on RHEL (or a clone of it) having more chance of solving a stability problem - its been more thoroughly tested and is very widely used. Having said that there's nothing wrong with MDK, just if you can't get it to work your last resort should be RHEL/Centos not Windows.

Hi TK,

I actually went out and bought another powersupply to see if it was a power issue. After I installed the 500W powersupply I also installed kernel 2.6.11 from 2.6.8 and it seems to be stable so far. I figured that since the computer was still powered, the NICS were ping replying and the kernel was answering calls but not throwing daemons it might just be a matter of the HDD's turning off.

I threw a network monitor tool on my workstation and it'll page me if the system dies. It's very erratic so it's hard to guage when it goes down. In the event it crashes again, I'll try turning off ACPI and see if that does it.