Re: RAIDframe corruption

[manu%netbsd.org@localhost (Emmanuel Dreyfus)]

Greg Troxel <gdt%ir.bbn.com@localhost> wrote:

[fsck on RAID halves]
> Presumably you mean on the raid0a.  Did you create e.g a wd0j and wd1j
> that are the halves (minus the 64 block header), and run fsck
> individually on those?

Yes, I checked raid0a, wd0e and wd1e, they are all fine.

> > Should the disk content be exactly the same? Does it make sense to
> > compare the whole 500 GB for differences? 
> After the first 64 blocks, yes, they should be identical, and yes it's a
> good test.

cmp -l produces a 1 GB output, with roughtly 3 regions of change:
- 32 bit word in MBR at offset 441
- One byte at offset  1064977
- 64 MB region at offset 498074652673 with many changes

   441 237   0
   442 272   0
   443 237   0
   444 272   0
1064977   1   0
498074652673   0  57
498074652674   0 165
498074652675   0 163
498074652676   0 162
498074652678 145 160
498074652679 164 153
498074652680 143 147
498074652682 154 163
498074652683 144 142
498074652684  56 151
498138882045  62   0
498138882046  40   0
498138882047  60   0
498138882048  60   0

-- 
Emmanuel Dreyfus
http://hcpnet.free.fr/pubz
manu%netbsd.org@localhost