My original question was on an error I was getting while doing a 0level dump:
DUMP:(This should not happen)bread from /dev/rsd1c [block -###]: count
(etc, etc)
Thanks to all who responded, especially:
robin.landis@imail.exim.gov
klfergason@amoco.com
Jim Harmon jharmon@telecnnct.com
Marina.Daniels@ccd.tas.gov.au
David Mitchell davem@fdgroup.co.uk
Several solutions were offered, and I'll summarize here:
Use ufsdump (block device based)...due to problems restoring to a
different disk geometry.
--We're still on Sun OS 4.1.3, so I couldn't try this
The filesystem is being accessed while the backup is happening.
--There were several attempts, & we're certain this wasn't the problem
This usually implies filesystem corruption...you should unmount the disk
and run fsck on it
--We did, and it comes up clean.
Check for disk write error messages in /var/adm/messages
--Nothing in this file that shed any light on the problem.
I think your disk has just about had it. You could try a low level
format and restoring the data, to see if this fixes the problem first.
We agree wholeheartedly with the last item. And in spite of fsck coming out
clean, it looks like corruption of inodes in more than one file is to blame.
We are currently tar'g and moving files to isolate the bad ones.
Again, thanks for all of the responses!
Vickey Kalambakal
This archive was generated by hypermail 2.1.2 : Fri Sep 28 2001 - 23:12:00 CDT