I understand what you are saying about this not being an OpenBSD or a
raidframe problem. I will try that tool you pointed me to and see what it
says. Will it permanently mark the blocks as bad? If the worst happens I'm
going to have to rebuild the system, but I don't want it to use those
blocks again on rebuild. Will newfs simply take care of it for me?
Actually wd1 is the disk causing these problems, but wd0 is the drive
marked as failed. Likely due to a crash when I was trying to do a backup.
The problems started when I tried to do a backup. Then when the system
came backup I noticed that parity reconstruction was failing. So I checked
and noticed that wd0 was marked as failed. An attempted in-place
reconstruction brings me here. So as of right now I don't have a backup.
My raid device is carved up into a few partitions. I'm going to save as
much of my data as possible. I'm hoping that those blocks are on a
"system" partition like /usr. Does anyone know of a tool that will tell me
which partition those blocks are in?
Next time my raid will be hardware based. :)
Thanks for your help thus far.
Jeff Quast wrote:
On 7/17/06, Jason Murray <[EMAIL PROTECTED]> wrote:
<snip>
In case the output is not clear enough, there is an error reading
block numbers 11722176 through 111722303 on wd1. This is not an issue
with raidframe. This is an issue with your IDE disk. (or ide
controller, etc...)
Test the disk thoroughly using badblocks from the e2fstools port and I
am sure it will reproduce the exact same console output. and panic,
though a different backtrace.
That OpenbSD crashes when an ide disk fails to communicate properly is
not the fault of OpenBSD, and definitly not raidframe. OpenBSD actualy
tried to do you a favor and step down the communication speed ( /wd1:
transfer error, downgrading to Ultra-DMA mode 4 ), in case it were the
fault of the ide controller or what have you.
I think I can easily guess that this is why raidframe marked the disk
bad in the first place. You need to replace the disk with a fresh disk
of similar or greater geometry size, copy the disklabel onto the new
disk, and reconstruct. This is what raid is for. Unfortunatly for you,
your raid is both software and ide. This is why the kernel panics.
