I think the OP just needs enough data to convince Samsung that the drive is bad so he can RMA it. If the drive is at fault the SMART data should do it (especially since the first thing Samsung will do when they get the drive is query the SMART data...)
Since the drive is part of a RAID array I don't think you can trust programs such as badblocks or ddrescue to accurately map bad "blocks" on the drive. Since a logical device like /dev/sda1 would represent more than one physical disk badblocks might be able to tell you that the array is failing or degraded, but mapping particular bad sectors on an individual disk in an array would be a pretty nifty trick. Similarly ddrescue would tell you that it could not recover parts of the partition, but again trying to figure out which disk in an array is at fault might be difficult. Hopefully it's a RAID 5 or similar so no data was actually lost. If either of these tools can report faults to that level I'd love to be corrected -- it would have been useful to me in the past! The other advantage of querying smart data is that you can do that while the disk is online -- no need to boot with a manufacturer diagnostic CD or bring a production array down to copy it with ddrescue or do a desctructive write test with badblocks -w. That is important in the case of live data people are working with. (However if the disk is part of something like a RAID 1 or 5 then the OP can just pull the drive and do whatever tests he or Samsung wants on it while the array rebuilds onto a replacement...) Just my 2c. James -- To UNSUBSCRIBE, email to debian-user-requ...@lists.debian.org with a subject of "unsubscribe". Trouble? Contact listmas...@lists.debian.org
