Hi guys,

First of all, MHDD is an awesome tool. Secondly, I wish I knew better how to use it.

I have a Toshiba sata drive that I'm running a full scan on, with Remap enabled. First I imaged the drive and verified the image. Then, because I had the image, I skipped the Erase function and went right to Remap. Was that correct, or is Erase necessary first?

I let the test run overnight, and this morning I found a UNC at 79% of the scan. I had checked "Spin-down drive after test", and maybe that's why MHDD was unresponsive? I could do nothing with it, and when I hit Alt the video began to break up.

So I rebooted and continued the Scan, same parameters except for spinning down afterward, starting at about 78% of the drive. I expected to hit the UNC again and see what I could do with it, but to my surprise there now was no UNC. So I don't know what's going on or what to do about it.

The two scans together produced 855 sectors of "<50ms" and 2 of "<150ms", and nothing worse except for the lone UNC. The drive spins at 4200 rpm.

The drive's SMART status was Healthy. Attribute 5 (Reallocated Sector Count) was 35, 100, 100, 50, which I believe is good. Almost all the normalized Smart values were 100, so am I correct that Toshiba uses that rather than 253 as the ideal state?

Thanks much for any insight you can provide.

Just erase that drive and rescan, should be good now.

_________________
www.datarecoveryne.com

Last night I cloned an OS onto the drive, but I never did Erase it first. Haven't tested it yet, but I've not seen any sign of the UNCs.

However, on another drive I've come across several UNCs, and unlike the first drive they have been persistent. I have no choice but to partition around them, other than tossing the drive.

Thanks,
p.

@paul1149,


That result is not surprising, if the cloning process wrote to the previously UNC sector(s).


I don't know if you care about that drive, or if that was just a passing comment about a situation that doesn't concern you. Truly persistent UNCs (which are different from UNCs which reappear after some time) are unusual on drives with much life left, in my experience. You can may want to review the SMART data for that drive.

From your previous post about the Toshiba 200GB drive:


I don't currently have a compatible x86 system to run MHDD and see what the column headings are for those 4 values - I'll guess threshold, current, worst, raw, but that guess could easily be wrong - but since none of the 4 values you gave for that SMART attribute is zero, then I would be concerned, if it was my drive. If you supply the full SMART info, inc raw values (e.g. from MHDD) then further interpretation may be possible.

That smart data was taken from a different program and is raw, current, worst, threshold.

I just partitioned around the second drive, and after repartitioning I'm getting some clicking still. I think this one is too far gone to play with. I did learn a bit from doing so, however.

Thanks,
p.

Thanks, so you likely have 35 reallocated sectors (i.e. the Raw attribute value) on that Toshiba disk. Based on my experience (and since confirmed by Google's data centre study), such a disk is significantly more likely to fail. Therefore I would be concerned for its longevity (and make extra sure I had sufficient backups etc.), although other people might not be concerned.

Thanks. Doesn't the normalized parameters, current and worst, being at the top of their game (again, I'm assuming that 100 is the way Toshiba calibrates the ideal state), indicate that the drive is still in safe territory (just as the overall SMART reads)? Or is you conclusion reading between the lines of the headline Smart judgment?

Yes - based on experience and the Google study which I mentioned before:

http://research.google.com/archive/disk_failures.pdf

Remember that you can't rely on SMART to give you any warning whatsoever, so looking only at the "cooked" (normalised) values removes a valuable source of information from consideration IMHO. You don't know what algorithm the manufacturer chose to calculate the cooked value. You also don't know the underlying cause of the reallocations which are going on, to be able to consider whether that problem is likely to accelerate. I suggest you read that Google study and the strong correlation of non-zero raw values of Reallocated Sector count with premature disk failure. They aren't making that up - I've seen that correlation too, but of course you might be lucky and in any case, it's your disk, so it's your choice what you do and I'm not trying to convince you either way.

Thanks again. Interesting study. With a finding like this:


...one wonders about Smart's value altogether.

OTOH, I see that even one scan failure indicates an increased likelihood of drive failure. Not good. But the data only went out eight months, and the curves had been flattening after the first two month "infant mortality" period. Makes me wonder about the ethics of including such a drive in a computer for resale.

FWIW, my view based on years of experience doing engineering work with large disk arrays (and hence many disks!), is that having SMART to look for predictive failure indications, is better than not having it. However its limitations (and design goals) are rarely investigated deeply by end users, which can lead to it not meeting their (unfortunately unrealistic) expectations, as I've had to explain many, many times in my day job. There is more to SMART than just failure prediction, but that's a whole different topic.

Good luck with your Toshiba disk. Hopefully it won't get worse, but if I were you, as a minimum I would regularly look for increases in the raw values of pending or reallocated sectors, and treat that as an urgent warning.

Thanks yet again. The problem is I'm refurbing and selling computers, and with that I do a bit of switching out drives, including getting used drives and taking bad sectors offline so they pass the tests. Now I find out that the industry standard is rather hollow. But then, the failure rate of new drives these days is pretty high as well. I guess I'll use a more pessimistic evaluation of the drives from now on, but I don't think I'll demand pristine condition.