WD Red 3TB hdd hangs on a bad section of the disk

[tahoebill]

I have a 3TB WD Red hdd recently marked as "degraded" by my Synology NAS. I placed the drive in an eSATA dock and formatted it in Windows and found that it passed "CHKDSK /x /f /r". When I tried to test the drive in Spinrite 6.1 using levels 1 through 3, it hung up in a loop trying to read a bad area on the disk, constantly switching between 16 and 256 sector reads.

 

[Steve]

I have a 3TB WD Red hdd recently marked as "degraded" by my Synology NAS. I placed the drive in an eSATA dock and formatted it in Windows and found that it passed "CHKDSK /x /f /r". When I tried to test the drive in Spinrite 6.1 using levels 1 through 3, it hung up in a loop trying to read a bad area on the disk, constantly switching between 16 and 256 sector reads. I wonder if Spinrite depends on the drive generating a timeout on unsuccessful reads. Normal desktop drives generate read timeouts, whereas NAS drives do not.

I think if you watch SpinRite's “Real Time Activities” screen you'll find that SpinRite is slowly moving forward rather than looping. If the drive has a "region" of trouble, SpinRite will gradually work its way forward through that region. During SpinRite 6.1's extensive development work we found that asking damaged drives to read larger blocks at once was far less useful than asking for only a few. So, when SpinRite encounters a problem is drops down to much smaller read requests while it works past the trouble. After encountering a problem with a sector it will drop to 16 sector transfers (as you saw) in the hope of getting that to succeed. If it can it'll cautiously move up to 256 sector requests for a while, then finally return to its highest-speed 32,768 sector transfers.

So, @tahoebill, from your description it appears that everything is working correctly.

 

[tahoebill]

OK, that seems reasonable. I just got frustrated watching it through over 12,400 iterations with nothing changing on the screen except switching the read size between 16 and 256 sectors. Maybe once the sector has irrecoverable bytes, the whole sector can just be marked as bad, rather than trying to read the rest of the bytes?

 

[Steve]

The trouble is, many users probably really really want to get every last bit of data from the drive. We DO allow SpinRite's data recovery "strength" to be tuned up or down — even down to zero! — but we've kept that as a command-line option since it needs to be used with care!

 

[PHolder]

So there is one thing you need to be aware of. RED drives are specially designed for use in a NAS, and run a different firmware that is supposed to allow the NAS to have more control. If you look on your drive label, it will probably say something like NASWare 2.0 or something. Because NASes are focused on data security, they are usually much more inclined to report a problem as soon as possible, to give you enough time to replace the drive before it fails completely. When it comes to NASes it's always better to replace a failing drive and rebuild your RAID before the drive dies, the RAID degrades or fails, and then your data is much more at risk. (You ARE doing backups though, right? RAID is NOT a backup.)