[Oneota]

Yes, I know this could go in a better forum. But no one visits the other forums on a Saturday night, and this is kind of important to get done soon, so I'm asking it here. Deal with it.

Anyway, we've got an XServe G5 with a 3-drive Striped RAID. Drive 3 has decided to die on us, and we're between backup strategies at the moment (literally - this drive chose the one weekend where we didn't have backups running to die), so we need to somehow scrape 2 files off that won't otherwise copy. Miraculously, we've gotten every other file off the drive except these two.

They're just regular data files, not executables, so there's no special worries about binary data, which means I'm open to any tricks you guys might have up your sleeves for reading data files in clever ways.

What I've tried thus far is doing a "cat filename > newfilename" which gets some of the file into a new one, but there's eventually an I/O error that stops cat in its tracks. I'm guessing there's just one bit of the file that's unreadable, so if there are any tricks like that that'll happily continue on in the face of a disk I/O error, I'd like to hear 'em.

Thanks!

[KeriVit]

Start Over?

[ibookuser2]

Use dd.

Code:
dd if=filename of=newfilename conv=noerror,sync

The bytes that cannot be read will be replaced with zeros.

[Oneota]

Originally Posted by ibookuser2

Use dd.

Code:
dd if=filename of=newfilename conv=noerror,sync

The bytes that cannot be read will be replaced with zeros.

Awesome! Exactly what I was looking for. Thank you!

I've got the server running DiskWarrior overnight (wishful thinking on my part, probably, hoping that it'll just magically move the affected files to another chunk of disk), but I'll try that in the morning.

Thank you again!

(In looking at the man page for dd, would it make sense to also use the arguments "ibs=1 obs=1" to minimize the amount of data that might be written with a 0 instead of the actual data?)