[stevelup]

Sandy, yes - I thought there were two of them!

[Pete L]

If you can't get a good read off a single sector I'd normally suspect the controller. But with both drives exhibiting the same symptoms I'm more bemused.

Don't know how a modern platter is encoded - is the timing information in the track itself or is there still a timing region? Again if you'd lost the timing off one disk it would be unlikely to hit both simultaneously.

How raw a read were you doing?

[stevelup]

ddrescue on a Linux box

The drives enumerate OK, but any reads just result in an IO error ( DRDY ERR ). I tried random sectors throughout the whole disk.

[Pete L]

Looking at a random 1TB WD Caviar Green I have in front of me (legacy of my own NAS failure ), I notice that the controller board is only secured by a four torx screws and has pad contacts to the drive assembly. No cabling needed to reassemble.

That suggests that if the fault is the controller, repair should be possible.

Note to self: must put drives into new NAS.

[stevelup]

The problem is I'd need to find an identical drive, plus I believe various coefficients and other bits and bobs are stored in the flash on the board.

In principle you could just swap it over, but as I understand it, it's not that simple.

[Pete L]

Did they come out of one of these Linux NAS?

Still feels like software to me if both drives behave the same. But I've not done any Linux driver debugging so I've got no advice for getting some clue as to what the interface is actually telling you. (In a former existence I did write device drivers but not for anything that isn't in a museum apart from the odd Satnav).

[stevelup]

It was a QNAP NAS.

As previously explained, the PSU went faulty (it was outputting 34V instead of 12V) which has, I believe, damaged or destroyed the drive controllers.

The error message simply means the drive is not ready (it never becomes ready no matter how long you leave it).

[Tall_Guy_In_a_PA28]

Once you let the smoke out of the drive electronics it is hard to get it back in again.

[dhs]

2) RAID is not a backup

Is that because of the way it was configured (RAID 0 / RAID 1 etc), or because a single point of failure failed, or something else?

You need different types of backup, depending on what type of failure you want to protect against. In the bad old days, the most likely failure was the spinning rust, so simple disk mirroring was an easy win. But, as has been observed, that doesn't protect you from a rogue power supply frying both disks, or the building burning down. For that, you need to copy the data somewhere else. But even that doesn't protect you against corruption of the underlying data (hence multiple copies)... So, you may need more than one backup mechanism, depending on the risks you are trying to protect against. "As I said: it depends..."

[Mr Bags]

I've had two major hard drive fails, one with pre back-up work data on (salutary lesson learnedl!) and it's loss made me almost physically sick. I thought there was no way I would ever get the data back, or even my client, but I was told about Kroll Ontrack in Epsom and they saved my bacon. Can't recommend them enough. I think they are now just called Ontrack. Good luck!

[stevelup]

Excellent - thanks. It's good to have first hand recommendations.