Same Defect in Multiple Drives

[Dr. Maybe]

Quick question out of pure curiousity:

We're poor scientists so we buy used SAS server drives in bulk as near-line backup and sanatize/check them with a FORMAT UNIT as they come in. We very often see that drives of a given model will hang during the format at the same spot.

For instance, we're currently going through 24 Seagate Constellation ES drives and 8 of the them have hung during the format at around 5.75%. They'll usually hang there for 10-15 minutes and then eventually finsh the format and run fine.

Why does this happen at the same spot across different drives? I assume that they're doing a remap at this point which would suggest a common manufacturing error or something in the servo programming that throws them off but . . . ???

Can anybody enlighten me on this?

Thanks.

[Dr. Maybe]

"Maybe" your SAS controller is doing something to the drives at that point ?

Interesting point. I'll run a couple using an HBA with a different chipset and see what happens.

Are you sure all the drives are OK and are you observing this only with a specific brand / model of drives ?

All of our data files are hashed with SHA-256 as they're written. We then verify the backup media by spinning up the drive every six months and re-generating/comparing the hashes for all files. We've yet to see a failure in any of the drives that have made it through a format when they first came in. The drives are offline in storage between test cycles so maybe they would fail more often if we kept them spinning but our experience is that if they complete the format and pass manufacturer's diagnostics they're good and stable for a long time.

As for manufacturers / models, my impression is that we see this much more often in batches of Seagate Constellation ES 2TB drives. I do not, however, have hard records on that. It may just be that we get a lot more of that particular model so it's just more noticeable.

Strange ... Not likely that all of those Seagates have defects on the exact same spot on the platter ...

Yeah, that's what's puzzling. The bulk of these drives are coming out of EMC or other large racks and we're reformatting them to 512 sectors. I thought maybe the specialty firmware might play a role but I just had one do it that's got the consumer firmware and was already formatted for 512 so that theory doesn't seem to work well either.

I'll push the next batch through using a different controller and see what happens.

Thanks for the help!

[ddrecovery]

I guess there is an outside chance that these drives were in use and suffered some sort of physical damage. Maybe someone moved the rack or something similar. I presume they were all set up as a RAID, so if this was just one RAID (not a SAN or multiple LUN's) then the heads would have impacted the platters at pretty much the same area of the drives and caused some slight media damage.

Total speculation but stranger things have happened.