Re: SCSI disk corruption

On Fri, Oct 17, 2003 at 09:27:33PM -0700, Michael Wolfson wrote:
> OK, so I've been trying to install the Sept 24, 2003 snapshot on my 
> non-Turbo mono cube.
> 
> I disklabeled it, got the bootloader installed (using Christian 
> Limpach's installboot.sh script), newfs'ed the partition, and copied 
> the GENERIC kernel over.
> 
> That boots.
> 
> The problem comes when I try to unextract the distribution sets.  I 
> ftp'ed them over to the hard drive.  Almost always (but not every time) 
> gzip -vt or tar -xzf fail blaming corruption of the file.  But I know 
> they're stored correctly on disk, since sometimes succeed.
> 
> pid 586 (gzip), uid 0: exited on signal 11 (core not dumped, err = 27)
> 
> What's odd is that gzip and tar have no problems if the .tgz files are 
> located on an NFS mount.  Also, cksum always returns the same value for 
> the .tgz files.
> 
> The other thing is that when I shut the system down, it always hangs 
> (or panics) when trying to unmount the root filesystem:
> syncinc disks... Done
> unmounting / (/dev/sd1a)...
> 
> Any suggestions?

Well, I just reinstalled my next to a new disk (the old one was too big, and
I couldn't close the box :).
I didn't notice any problems. I extracted pkgsrc.tar.gz to the local disk
(from both local disk and ftp -o -).
Did you use tar -v ? I didn't.

BTW, softdep filesystems seems to work fine :)

-- 
Manuel Bouyer <[email protected]>
     NetBSD: 24 ans d'experience feront toujours la difference
--