Hi Gang,

I've a hardware RAID5 that is pretty sick. I've gotten it up and mounted again, and I'm attempting to make a backup before proceeding with further fixing. The problem is clearly some of the data are corrupted. An example: in my home directory, a subdirectory "sequences" is listed in the tree. However, cd sequences, vi sequences, cat sequences, less sequences (you name it) gives "no such file or directory"

Is there a switch to cp (or some similar program) that will allow me to copy through read errors? I'm aware of the usefulness of the noerror option to dd in this regard, but I'm backing up on a volume of a different size than the RAID and I'm not sure how to handle that situation without having detailed knowledge of the usage pattern of the RAID itself.

Any insights would be greatly appreciated.

Thanks

Dave Morgan
Biochemistry Department
UTSW Medical Center at Dallas
david.morgan@utsouthwestern.edu

The "find" command might be what you want.

To get an idea of what the find command does, log in as root and then do this command:

find /

(Instead of /, use the mount point for your problematic RAID setup.)

I'm hoping that it continues all the way to the end for you. If it doesn't, my answer won't help you at all. If it does, you can use the find command in one of two ways.

The more straightforward way is to use the -exec option of the find command. Be sure to place

\;

as a separate item at the end of the -exec option. The backslash is necessary so that the shell will pass the ; to the find command, instead of interpreting the ; itself.

The trick is to create an -exec option to the find command that does the copying of each file or directory properly.

But if the -exec option doesn't give you what you need, try this:

find / > somefilename

Then use Perl or something to read that output file and create a (very long) shell script which does exactly what you want.

Hope this helps.