andreas wrote:
Am 20.07.2010 14:03, schrieb Chris Bennett:
andreas wrote:
Hello All,
I am new to this list and OpenBSD. I installed OpenBSD 4.7 i386 from CD
and FTP.
So which one is it? ftp or CD?
I used CD for boot and prepare, install from Swiss ftp mirror.
First I overseen Sorftraid, so I used CCD to mirror my hard
disks.
What are your bios settings for the disks?
The disks used for OpenBSD are set to singel drives in the bios of the
PCI-Sata SiI3114 card, no "fake" raid, no IDE emulation is set. On the
same 2 disks is also a linux with Softraid, without any noticeable
problems. Behind the on-board VIA VT6420 SATA is a winxp with bios based
fakeraid with 2 other disks.
After short use of the fresh installation I got a Segmentation
Fault in more and less. Badly I did not use gdb at this point, because
I assumed a failure on my side during installation or on my hardware.
Then very early I removed all and start again with Softraid. After a
few reboots, I was only getting familiar with OpenBSD and did some
little configuration work, sshd was not starting anymore.
What did you mess with to screw up sshd? You did an install, not an
upgrade?
I only use the option install, nerver update. After install to a singel
disk, I booted into singel user, backup all, prepare and create
softraid. Than after a couple of boots first time more/less was broken,
than later ssh/sshd (libcrypto.so.18.0)
Again I got
Segmentation Fault, this time from sshd and ssh. Now I used gdb, and it
was telling me about some problem with libcrypto.so.18.0, my bad I did
not keep this file, I directly overwrote it with the lib from my backup
, which I needed to create the Softraid. After this sshd and ssh was
working normal again.
I checked my disks and my ram, they are fine.
How did you check them? Hopefully not by using SMART.
Did you try swapping drive cables for new ones? I have had MANY drive
problems that were from bad cables
I used smartctl from linux side (because Im still more familiar with it,
and I used atactl. For the ram I used mem86test, overnight, without any
errors. Because other OS is working fine, I did not consider the cables,
but I give it a try.
How should I check the hard disks?
smartctl looks at SMART data, which is almost useless.
try adding e2fsprogs and use the program in there called badblocks.
You will want to use the non-destructive form of a read and write test.
This will actually check the partitions for "bad blocks". Read the
manual page carefully