{The complete intended title was : “Serious issues with a ST2000DM001, seemingly not-so-serious issues with a ST2000DL003”.}

Hello,

I have a ST3000DM001 HDD which seems to be seriously failing. Not so long ago it only had 16 bad sectors ; they appeared at the end of July, but the count didn't increase further, and I was already struggling to find free space to satisfy my data-hoarding tendencies, so I continued using it with no safety net (foolish, I know !). Then in mid-October I made a surface scan with HD Sentinel (read only), which increased the number of pending sectors to about 100, but allowed me to locate the bad areas – two small intervals near the 2TB mark. So I tried to identify which files could be affected, using various methods (I asked the question in the first linked thread, and wrote the long reply in the second), then I moved the identified files (6 large videos) to a special folder, and proceeded to copy the rest of the files and folders by order of importance to a healthy HDD (only one additional file could not be copied : an unimportant .eml file, which happened to be located near an already identified damaged area, based on data provided by nfi.exe and Defraggler). Then I tried to salvage those remaining video files, using ddrescue – and the HDD's state turned from problematic to catastrophic in a matter of hours : now the SMART data indicate more than 12000 reallocated sectors, plus about 9000 pending ones. And it makes that sound, almost constantly :
http://www.cjoint.com/c/GKxchyuW7a7
I can still start a copy with ddrescue, at least the begining can be read, if very slowly ; during such an attempt, the weird noises cease, then almost as soon as read operations are interrupted it gets noisy again. And the number of reallocated sectors keeps increasing, even when the HDD is running idle.
I didn't think that it would deteriorate so quickly. I wanted to make a thorough comparison with WinMerge to ensure that all the files were flawlessly copied, but now it seems to be too late for that. All the files I checked seem to be totally fine, though, and I guess that there would have been an obvious failure warning if some more sectors could not have been read (but I can never be sure – I already had cases of silent corruption).

Questions :

– Was it the right move in such a case to first isolate the files occupying bad sectors, and copy the rest by order of importance, as opposed to doing a full image with a tool like ddrescue ? Judging from the way it behaved when attempting to seriously access those bad areas, it's possible that I would have failed to complete the image, which would have left me with much more significant data loss (especially considering that a portion of the MFT was located near the very end, without it I wouldn't have been able to recover the complete directory structure, or to successfully recover fragmented files referenced there, or even to know exactly what was there in the first place).

– Is it common for a HDD to deteriorate so fast when trying to read bad sectors ?

– Are there some advanced tricks that I could use at this point to somehow be able to salvage some more sectors and try to complete the recovery of those damaged files ? Or does that sound and general behaviour indicate a severe hardware failure, like a damaged head ? I recently discovered a program called HDDSuperTool, which can send advanced commands to a HDD. Could it be useful here, and how ? (I tried reading the manual, but so far I'm kinda clueless about what each command does and when they could be used.) Those six video files (I had to improvize a sophisticated trick to actually recover them since I had made a partial image which didn't include the whole MFT before the HDD's state became too critical, and several of them were highly fragmented) are not super important, and I'm aware that I've been lucky to have only 6 corrupted files, running a 3TB HDD for months with no backup, but still, I'd like to see what can be done, now that the poor thing is definitely on the way out anyway, and the rest is already saved (well, not backed up yet !...).

– Is this particular range of Seagate HDDs known to have that kind of issues ? (I read several comments here expressing a very negative opinion toward Seagate HDDs in general.) Could it be related to heat issues ? (I saw a video by Scott Moulton saying that in some cases the pre-amplifier was merely fixed with glu to the head's chip, and could get disconnected because of excessive heat – if I understood this correctly. But my hard drives are well ventilated, and I check their temperature constantly with HD Sentinel, it never got too hot, even during the hottest days this summer.) I also had a ST2000DM001 which began to have hicups during this summer : disconnected itself randomly, caused a system crash, then HD Sentinel displayed a red warning, 280 bad sectors (strangely a few days before the health status had gone down to 12%, then back to 100%, so I thought that it was only a glitch, as Dick Jones would have said). I successfully cloned it using ddrescue (to a seemingly perfectly healthy ST2000DX001 – well, another Seagate), only 600KB were damaged near the 3GB mark (which belonged to $UsrJrnl, a NTFS system file, I don't know what its purpose is exactly, but at least no personal files were corrupted). I thought that I could still use it for non critical data, by creating a new partition beyond 4GB, but then its state deteriorated badly in a matter of days, as I was copying stuff on it.
And I have an older ST2000DL003 which happened to also have 280 reallocated sectors, plus 8 pending sectors. When I saw that, about a year ago, I copied everything (apparently without issue) to another 2TB HDD, then I hadn't touched it since. A few days ago, after my troubles with the ST3000DM001, I plugged this one, and made a thorough comparison with the backup using WinMerge : all the files and folders were found identical – and the 8 pending sectors disappeared in the process (its health status in HD Sentinel went from 20% to 23%). Now I'm performing a read surface scan with HD Sentinel. If the scan turns out 100% clean, with all green blocks and no slow-down, can I safely use it again, or is it bound to follow the same trend as the ST2000DM001 and ST3000DM001 ?
Update : It's all green, no slow-down.

– Regarding the ST2000DL003 again, as there are 280 already reallocated sectors, is there a way to know where they are / were located ? As far as I know, reallocation can only happen when data is being written, in principle, but my experience with the ST3000DM001 proves that it's not always the case. So how can I be sure that no file was corrupted ? And could it have been a temporary issue, just like the 8 pending sectors which disappeared after a successful read attempt, or rather were considered good again ? Is there a way to clear / reset the list of reallocated sectors so as to check them again, with free software tools, or does it necessarily involve special interfaces, very expensive and meant for professional use, like the apparently staple PC3000 which is regularly mentioned here ? (280 sectors isn't that much, space-wise, but it's the difference between a perfectly operating HDD and a suspicious one.)

I must say that I consider myself pretty well versed when it comes to the specific field of data recovery, I've helped several persons with relatively severe hard drive failures, each time getting excellent results so far, but reading the replies from true experts on a forum like this I feel quite humbled ! :^p I mean, the knowledge gap between a serious amateur and a bona fide professional seems to be staggering in that particular “sector” of activity... Yet I'm trying my best to improve my modest knowledge base, so as to deal as efficiently as possible with my own SNAFUs, provide some help when I can, and at least not screw things up any further when it appears that I can't...

Best regards,
Gabriel


PS : I got a warning when trying to add the audio recording as attachment that “The extension mp3 is not allowed” – yet that would be kinda useful here... The file should be accessible through the cjoint.com link above, but I cannot guarantee how long it will stay there, in case this thread gets highly informative replies, if someone reading it in the future and having a HDD with a potentially similar issue wants to check what it sounds like and if those informations could be useful in their situation.

Update : Now the ST2000DL003 has 288 reallocated sectors, so the 8 previously pending then considered good again were apparently not-so-good after all... provided that those are indeed the ones which have been newly reallocated ! What's frustrating is to not even be allowed to know where those problematic areas are located, if they are scattered all over the surface or all in one particular spot, that would help in deciding whether a hard disk drive is still fit for a relatively reliable use or if it has to be put out of duty right away (but then again, see above, I thought that using the ST2000DM001 would be safe, considering that all its bad sectors were in the same place, yet it proved to be highly unreliable).
(And by the way I don't quite understand how HD Sentinel calculates the health ratio : with 280 reallocated sectors + 8 pending ones it indicated 20%, now with 288 reallocated sectors it's 22%, as if it were somehow better...)

Well... was that that pointless, or uninteresting, or too long, or what ?

Hello,

backup data and replace the drive with some other brand.

Well... I was expecting something more... substantiated, coming here... Some technical input to either help me perform some advanced tricks on those failing HDDs (in an attempt to recover those damaged files, or at least to learn something useful), or explain why it's irrelevant, or why nothing more can be done without serious stuff I can't afford.

Some other brand – meaning Western Digital ? Considering that there are only three HDD manufacturers worldwide these days (since HGST now belongs to WD, and Samsung's storage division now belongs to Seagate), while Toshiba's RMA policy is practically non-existant for internal 3.5" HDDs, at least in Europe.

Anyway, I don't really understand what point you're making here : any hard disk drive can fail at any moment, apart from specific series with an abnormally high failure rate, the best one can do is backup regularly and hope for the best. I've had several WD HDDs failing, one Toshiba (only one I bought, probably the last for the reason mentioned above – unless they implement a decent RMA policy for 3.5" internal HDDs in Europe), I've recently recovered data for someone from a 1TB Hitachi which failed suddenly for no apparent reason, and quite badly (yet that brand seems to be held in high esteem here), so far I had had no trouble with Seagate, now I've got several of them failing in short succession : granted, that's not reassuring, but that's not statistically significant. For instance, the ST2000DM001, I bought it used 3-4 years ago, along with two WD30EZRX : the two WD were DOA, there were actual signs of shocks, the seller didn't want to acknowledge the defects, but fortunately I could get two functional ones through RMA ; the Seagate also had a small bump, but was perfectly operational, and after thoroughly testing it I've been using it as my main HDD until this summer : I'd say 3-4 years is not so bad for a HDD these days (and its performance has been consistently excellent). The ST3000DM001, I've also bought it used, but only a few months ago, although its production date is 13263 so it's from 2013 – not so bad either if it's been used regularly since then.
Reading this forum regularly for about half a month now, I've noticed that Seagate drives, and particularly those with “F3 architecture”, had a bad reputation, but I still haven't found a clear explanation as to why exactly they are so worse than their counterparts from other brands, why they are supposedly more prone to failing, and/or failing in a worse way, making recovery attempts more tedious. Do you have serious statistics to back up such statements ? (Other than the dubious Backblaze reports.) Or could you provide some technical insight, worthy of that forum's name ?

Data recovery professionals deal with hard drives every day. We know what drives are good and which have inherent issues. Seagate drives are very prone to head failure. Unfortunately when the heads fail, they quite often cause damage to the media surface making them unrecoverable. We see it every day, especially the 3TB DM drive. WD's are much more forgiving both in terms of the types of failure and the chance of recovery.

Without professional data recovery tools it is pretty much impossible to recover hard drives with any significant failure such as firmware or head failure. The drives you have successfully recovered were probably not significant failures. I am sure you thought they were, but perhaps the issues were not too severe. I am not trying to downplay your talents and your eagerness to learn, but without professional tools you stand the chance or making hard drives you work on unrecoverable by not diagnosing the issue correctly.

_________________
HDD, SSD, Flash and RAID Data Recovery
Founder of The Data Recovery Professionals Group

@ddrecovery
Thanks for these informations. At least that was more helpful than just “replace the drive”.

Some more questions (or remarks) if you don't mind :

1 – I have another ST3000DM001, currently perfectly operational, which I also bought used a few months ago, more recent but out of warranty. Should I re-sell it based on the knowledge of that presumed high failure rate ? Or would you consider this to be a dishonest gesture ? Should I add a disclaimer in the anouncement saying that this is a reputed crappy model and anybody buying it should expect it to fail miserably within a year if they're lucky ? Or should I accept my fate, keep it and deal myself with the burden of knowledge that it is bound to fail ? (My cat has had a mammary carcinoma for two years now, and is doing really fine considering, so I'm getting used to preparing myself for the worst...)

2 – I am not trying to downplay your expertise, but, since you (DR professionals in general) only deal with failing hard disk drives, how can you make general statements regarding a brand's reliability, or lack thereof, for their entire range of models, based on your daily experience ? I have no interest in saying that Seagate HDDs are good, I'm just pointing out that such statements should be more substantiated than what I've been reading on this forum. If you have more Seagate HDDs coming your way, it could be that more units of those are sold, it could be that they're more vulnerable to heat and you're located in a hot area, or any other reason. The potential for a successful recovery is a different matter, and on that aspect I'll have to trust your judgement, but nobody wants to end up having to pay a high fee for a data recovery, so that aspect is not taken into consideration when buying a particular model, except maybe for large corporations. As a personal user, the only relevant criterions are the actual average failure rate (which is very difficult to assess with enough statistical confidence), and the reliability of the RMA service during the warranty period (I have yet to make a request for a Seagate unit, since none failed within the warranty period, otherwise I know that WD's service is top notch but they only provide “recertified” units as replacement, which is not very reassuring, and Toshiba does not provide a direct RMA service for 3.5" internal models). Again, any HDD can fail and the only safety measure is having backup(s).

3 – Again, that's not statistically significant, but when those two Seagate ST...DM001 I mentioned failed this summer, I could recover almost 100% of their content. On the other hand, I've had a WD6400AAKS (bought new in 2009) which failed too, but it began clicking right away, for no reason except perhaps prolonged use in a non ventilated external enclosure during the hot days of summer (SMART report was perfectly clean up to the moment it made a weird noise when turning it off). I haven't tried to plug it since then to attempt a recovery, as the data it contains is not that important, but my hopes are rather low for this one. If it turns out that it really failed to the point that it would require a professional intervention, then it contradicts your statement that WD HDDs are more forgiving in general (as I will definitely not bring that one to a professional DR company, so if that turns out to be the case the recovery ratio is going to be 0%).

4 – Obviously, at this point I can only do software recovery, so I'm well aware that if I can do something at all, the failure is not so severe (yet) that it requires hardware intervention. Even then, when I propose such a service (so far mostly for friends / family, only had one real “client”, who paid 50€ for a near total recovery on that aforementioned 1TB HGST HDD, which was clicking all over the place and very unstable, with the ddrescue + R-Studio combo, call me lucky – yet he was dumb enough that he couldn't find his personal files among the complete Windows directory tree...), I always stress that it's risky, that any attempt at powering up or accessing a failing HDD can further damage it, and lower the chances of recovery if a professional DR company is ultimately requested to work on it. Some people just can't afford such a service, or could with some effort but do not want and are willing to take a chance – hell, some people are even willing to try insane things on their own, you have some of them coming here once in a while ! :^p (For instance, I've had a request from a guy who said his laptop HDD was having trouble, he said that he considered himself well-versed with computers but didn't know how to deal with that. I asked him a few questions, including if he had seen SMART data lately, but insisted that he should not power it up now just for that purpose... yet, in his next reply he said that in the meantime he had been running... CHKDSK... for... 24... hours... and then, guess what, the poor thing completely stopped responding... Another dude is looking for a HDD just like the one his daughter dropped on the floor, in an attempt to transfer the “two disks” inside, meaning that he already opened the poor thing ; that one says that he knows his ways with car engines and whatnot, and thinks that anything can be done with enough perseverance, that people just don't dare to open HDDs ; he said he had already dissassembled another unit with “one disk” which was given to him, and it still worked afterwards. I tried to explain why it had virtually no chance of working, but he was stubborn, so I wished him good luck in his endeavour.)

5 – How can a firmware failure happen spontaneously ?

6 – In the present request, I asked if it was possible to attempt something out of the ordinary with my own drives, in which case there's nothing to lose, as almost 100% of the data could be recovered, and the remaining damaged files (six MP4 files) could probably not be recovered in full even by a professional. I wouldn't attempt some untested trick on a drive that isn't mine.

7 – So, in this particular case, from the behaviour pattern I described and the sound sample I provided, what would be your diagnosis, based on your experience with that particular model ?

8 – The ST2000DL003 has had some more reallocated sectors, total 344 now, even though I did not write anything on it lately (and surface scans still come out clean). How can this be explained ?

Drives perform "background media scans".

Savvio 15K SAS Product Manual, Rev. C:
http://www.seagate.com/staticfiles/support/disc/manuals/enterprise/savvio/Savvio%2015K.1/SAS/100407739c.pdf



I believe Seagate's replacements are usually recertified. As for reliability statistics, Backblaze is very often cited as a good indicator in this regard.

https://www.backblaze.com/blog/hard-drive-failure-rates-q3-2017/


The drive's grown defect list {G-list) may overflow.

WD drives frequently develop problems with the "relo-list".

Seagate drives suffer from media cache faults and the "pending bug".

Samsung drives suffer from corruption to the "FIPS" module in the ROM, resulting in a no-spin situation. Some Samsung models have a firmware bug which bricks the drive when PUIS is enabled (as was the case when Windows 10 was first released).

SMART data and defect information are written to the platters. If a head begins to fail, then any SA writes by this head cannot be trusted.

If you intend to use ddrescue to recover your data, at least try to disable reallocation and retries. This will reduce the stress on the drive.

http://www.hddoracle.com/viewtopic.php?f=59&t=1842

_________________
A backup a day keeps DR away.

"Data recovery" not only deals with failed disks, they also deal with damaged disks (burnt, watered, ...), and also with perfectly fine working disks (data wiping). The damaged and perfectly fine disks they receive should be somewhat representative for the disks sold on the market in geographic region of the customers, while I agree that the failed disks are likely biased.

Actually the idea is to backup properly so you don't need to worry about a fail.

As for DR Pros knowing quality.. when you work on drives you also get an idea of quality.. Like when I worked on my "cheap chinese" Quad bike, I could easily tell the alloy parts were shit, threads readily stripped, cables were weak etc etc, and I am guessing a Kawasaki or Suzuki would not "feel" this way.

How can firmware fail "suddenly"? well you forget it isn't just sitting there doing "nothing" there is a lot going on, and we all know nearly all software has bugs.

When you have millions and millions of a certain product, especially hard disks that need to be almost perfect 100% of the time, but are treated to many environments and use cases... you are going to get a list of issues that run the gamut.. micro analysing a minute subset of it is frankly ridiculous.

I certainly cannot advise you on how to sell your drive. There is a class action lawsuit out against Seagate for continuing to sell drives with such a high failure rate.


As DR professionals we deal with a lot of IT techs either from IT companies, tech support or computer stores. We see and talk to them every day. They have experience of both good and bad drives and pretty much all of my clients do not use Seagate drives anymore.


You could buy 10 of the cheapest cars on the lot and they might all run for 5 years without an issue. You could also buy a Ferrari or a Porsche and it breaks down on the first day you drive it. Hard drives like cars are mechanical and are therefore prone to failure. You are making comments about hard drives from very limited experience of them and presume that everyones experience must be the same. We talk to people like you everyday and we get a feel for good and bad drives.


That could be a prime example of not understanding or diagnosing the failure. If the client couldn't find his personal files they probably were not recovered. A few bad sectors on the wrong place in the Master File Table would mean the complete folder structure may not have been recovered.

_________________
HDD, SSD, Flash and RAID Data Recovery
Founder of The Data Recovery Professionals Group