0

Today, I have gone through the process of individually plugging in all my ten current backup disks and checked them with CrystalDiskInfo 8.17.8 x64 in order to determine whether I dare to keep using them or if they are "singing on their last verse", so to speak.

All are ~10 years old and are reported as "Good", except for one which is labeled "Bad". I dumped the report data for that one.

But this "Bad" disk appears to work in practice. It does not do that typical, awful "hangs trying to access the disk forever" when I open it in File Explorer, and I'm able to transfer a bunch of big video files out from it to C: without issues. So what does it mean that it's "Bad"? And can I really trust that my other ones are really "Good", if this "Bad" one is actually good?

Also, I find it curious that it claims to have been powered on for only 1 hour. This makes no sense since I know for a fact that this cannot be accurate, and all the other disks of the same age and usage report like 100 hours or something. But in the other section of the report, it says "100 100 __0 00000000005F Power-On Hours", which looks like 100 (not 1) hours to me? What is that about?

That alone makes me doubt all the other data, but I'm not an expert on this, so I will leave the deep analysis to those of you who are:

----------------------------------------------------------------------------
 (05) WDC WD10TMVW-11ZSMS5
----------------------------------------------------------------------------
       Enclosure : WD Elements 1042 USB Device (V=1058, P=1042, sa1)
           Model : WDC WD10TMVW-11ZSMS5
        Firmware : 01.01A01
   Serial Number : [redacted for privacy]
       Disk Size : 1000,2 GB (8,4/137,4/1000,2/1000,1)
     Buffer Size : 8192 KB
     Queue Depth : 32
    # of Sectors : 1953525168
   Rotation Rate : 5200 RPM
       Interface : USB (Serial ATA)
   Major Version : ATA8-ACS
   Minor Version : ----
   Transfer Mode : ---- | SATA/300
  Power On Hours : 1 hours
  Power On Count : 49 count
     Temperature : 24 C (75 F)
   Health Status : Bad
        Features : S.M.A.R.T., APM, NCQ, GPL
       APM Level : 0080h [ON]
       AAM Level : ----
    Drive Letter : G:

-- S.M.A.R.T. --------------------------------------------------------------
ID Cur Wor Thr RawValues(6) Attribute Name
01 __1 __1 _51 000000001CF4 Read Error Rate
03 161 160 _21 000000001356 Spin-Up Time
04 100 100 __0 000000000038 Start/Stop Count
05 200 200 140 000000000000 Reallocated Sectors Count
07 200 200 __0 000000000003 Seek Error Rate
09 100 100 __0 00000000005F Power-On Hours
0A 100 253 __0 000000000000 Spin Retry Count
0B 100 253 __0 000000000000 Recalibration Retries
0C 100 100 __0 000000000031 Power Cycle Count
C0 200 200 __0 00000000000B Power-off Retract Count
C1 200 200 __0 000000000066 Load/Unload Cycle Count
C2 126 104 __0 000000000018 Temperature
C4 200 200 __0 000000000000 Reallocation Event Count
C5 200 197 __0 000000000021 Current Pending Sector Count
C6 100 253 __0 000000000000 Uncorrectable Sector Count
C7 200 200 __0 000000000000 UltraDMA CRC Error Count
C8 100 253 __0 000000000000 Write Error Rate

I suppose "Bad" probably means "Bad situation" rather than "Bad drive", so I will probably have to buy a replacement disk (yet again...) even though I've not saved up enough for hardware to die this frequently...

image version of the report

Whigham
  • 1
  • 2
  • What S.M.A.R.T attribute exactly is the tool calling out as the reason for the drive's status. I assume its the current value of `Read Error Rate` but that would only be a guess. 5F is 95 hours, I would assume the tool, is to blame for displaying the incorrect value of hours it's been turned on. – Ramhound Oct 21 '22 at 13:39
  • @Ramhound I have updated my question with a screenshot. – Whigham Oct 21 '22 at 13:49
  • 1
    How you read `Read Error Rate` is that the current attribute value is 1, and the threshold to report an issue was 51, and the current number of the read error rate is 7,412. In other words the rate of read error rate is so high the drive likely has numerous bad sectors, so while data is being written to sectors on the drive good luck reading that data back. – Ramhound Oct 21 '22 at 13:54

1 Answers1

0

SMART data is only an index whose value is meaningful only if interpreted based on the manufacturer's guidelines, since the standards of implementation vary between manufacturers.

As far as simply taking these values at face value, in all respects the SMART values on this 10-years old disk do not make any sense.

There are no Reallocated Sectors or Spin Retry, Seek Error rates are low, Power-On Hours are 95 in 10 years while being Power Cycled only 49 times etc.

In short, I don't believe that these numbers are correct, which doesn't inspire in me any confidence in this disk.

The "Bad" label was probably inspired by Read Error Rate being 7412. This parameter indicates the rate of hardware read errors that occurred. Any value differing from zero signals a problem with the disk surface or read/write heads. The higher parameter’s value is, the more the hard disk failure is possible.

The Current Pending Sector Count is also 31, which indicates the number of unstable sectors on the disk drive that haven't been managed to remap or reallocate.

In short, errors are indicated, even if the values look somewhat funny, so you would do wisely to not trust this disk in the long term, even if it works for the moment.

References :

harrymc
  • 455,459
  • 31
  • 526
  • 924
  • Note that for long-term storage, hard disks, even if not used much, should have their data refreshed every few years, since magnetism is not eternal. – harrymc Oct 21 '22 at 13:50
  • More specifically the threshold of that attribute was reached, since it's currently 1, instead of being above 51. – Ramhound Oct 21 '22 at 13:55
  • @harrymc How to "refresh" the data? – Whigham Oct 21 '22 at 14:43
  • Copy data away, reformat the disk using slow format (not quick) or manufacturer's low-level format, then copy it back. The disk and data would then be in great shape. There exist longer-term solutions like M-Disc and Blu-Ray, but they are costly. – harrymc Oct 21 '22 at 15:05
  • @harrymc This is the very first time I hear about this in my life, and I have been actively looking for exactly this kind of information... Are you sure giving it power every few years isn't enough? – Whigham Oct 21 '22 at 15:19
  • Not for an HDD, since unlike an SSD the firmware is not testing the disk repeatedly. An HDD, even on the shelf, would suffer from some amount of bit-rot. Note: Add to my previous comment that after all this copying, you need to examine the SMART attributes of the disk before letting it lie for a few more years. – harrymc Oct 21 '22 at 15:23
  • Please tell me there is a tool that does this automatically if you let it loose on a mounted drive, so I don't have to actually sit and move all the files over to my local disks and then move it back again? (A tool which does this properly and safely, file by file.) – Whigham Oct 21 '22 at 15:25
  • There are disk-copying tools that would do that on the disk-level, but they are also not cheap. The most you can do is create a script and let it do the job unattended. – harrymc Oct 21 '22 at 15:27
  • 1) The OP talked about one of his ten backup disks: if only plugged for backup from time to time, the displayed values are plausible (95 hours powered on in 10 years would be around 1h per month... I have a disk that is not powered much more than that). 2) the normalized values don't required manufacturer guideline, and the one for read error rate is well below the threshold: this *is* bad... – PierU Oct 21 '22 at 16:16
  • @PierU: This is a 1 TB disk, 5400 RPM, which should have taken some hours to even copy. Also, for a really bad disk there should have been some other errors such as Reallocated Sectors, which are completely missing. The ensemble is very puzzling. – harrymc Oct 21 '22 at 16:20
  • and 3) I'd really like to have a source about HDD demagnetization along time. Sure there is some amount of, but in practice it doubt it can be an issue for a drive that is stored in proper conditions. – PierU Oct 21 '22 at 16:20
  • I don't know how the OP is using this disk. If he's storing some files and never delete/replace them, then the reallocated sector count is also plausible. A defect sector can be effectively reallocated only at the moment something new is written on it. – PierU Oct 21 '22 at 16:25
  • @PierU: A failing disk whose data is mostly static over the years shouldn't have so many errors, and only errors of one kind. This is weird. – harrymc Oct 21 '22 at 16:31
  • To be clear: all of these disks spend 99.99% of their time in a safe and two of them (the oldest touched) get taken out and plugged in to sync data every two months or so. – Whigham Oct 21 '22 at 17:05
  • @Whigman: Proof for "low-level format" nonsense: "Low-Level" is a marketing term today. Read https://www.seagate.com/support/kb/how-do-i-low-level-format-a-sata-or-ata-ide-hard-drive-203931en/ As for harry's "There are disk-copying tools that would do that on the disk-level, but they are also not cheap." check out the linux dd command or even better, the ddrescue command (the one from GNU foundation) on linux. But for harry a price of zero dollars seems to be too much. – r2d3 Oct 21 '22 at 17:33
  • @Whigham so that confirms that the low powered-on time is not weird... About the data retention time I don't think you should worry too much, last time I read something about that it was after a quite long time that demagnetisation could be an issue (something like 20 years). If it was such an issue, the HDD controllers would take care of the problem (exactly the same way than the SSD controllers take care of the specific issues of the SSDs)... – PierU Oct 21 '22 at 19:59
  • ...However, after a long time (10 years is a long time) it would not harm to do something about that. The easiest way is to have a spare disk for the rotations, and you copy the whole content of the oldest backup to this disk. Not only the new copy is fresh (in terms of magnetisation), but this is also a verification that the old backup was still readable. Not need to low level format or whatever... – PierU Oct 21 '22 at 20:03
  • @Whigham: Now that this guy is gone, we can continue. I added 2 articles to my answer with interesting quotes. I think we can all conclude, whether we fully understand or not the SMART values, that your disk is weakening and cannot be trusted. Whether you decide to junk it, or use it for non-essential data, or only as secondary or tertiary or more backup, this is up to you. Be careful if have more of those 10-years old disks. – harrymc Oct 21 '22 at 20:13
  • In any case, powering the disks does not add to their longevity, is only useful for checking their SMART stats (or not even that). Refreshing from time to time *all* the data including the file-system tables, will verify through the SMART data what is the state of the disk. – harrymc Oct 21 '22 at 20:19
  • @harrymc I already saw the two references you are linking. The first one is full of at best mere and unsupported statements, poor interpretations of existing stats (Backblaze stats do show the the drives can last many years, but the author strangely concludes that they last 3 to 5 years), and at worst of technical nonsenses. Reading *"To ensure a longer lifespan for your hard drive, it is essential that you do not use it too frequently"* leaves me voiceless... The second one is more correct but doesn't deal at all with data retention. – PierU Oct 21 '22 at 21:31
  • @PierU: A disk's life is a matter of pure luck. The first reference was one extreme opinion against using hard disks as backup for decades. The second is more serious, and its 10-years number is to be taken seriously. Note however that only enterprise-level disks are rated *by the manufacturers* for 10 years, others for at most 5 years, no matter whether they are used or not. With good care, HDs may further extend their life-time, but there is no guarantee. For long-term storage, non-magnetic media is preferable. – harrymc Oct 22 '22 at 06:40
  • @harrymc As you note, the first reference is just made of opinions. It's a personal blog that would not be accepted as a valid source on Wikipedia for instance. In the second reference I can't see the "10 years" you mention, and anyway 1) it is based on Backblaze stats, which are very good ones but which are about drives experiencing a heavy workload 24h/24h during years. Even the drives in my NAS are nowhere near such a workload; 2) it has nothing to do with data retention on unused drives; 3) don't take warranty periods as a life expectancy. – PierU Oct 22 '22 at 12:45
  • @PierU: Manufacturer lifetime rating is worth a much as food expiration dates, meaning keep the manufacturer safe. But they do indicate whether the disk is built to be more or less robust. As I said, in the end all is a matter of luck. But whatever good care the disk receives, its lifetime is still limited. The numbers of 5-10 years keep cropping up and should not be totally ignored, and a disk older than 10 years is entering the danger zone. – harrymc Oct 22 '22 at 13:12
  • Manufacturer do not give anything like a "lifetime rating in years". They give a warranty period that is just "If the drive fails before x years we get refunded". A typical PC has a 2 years warranty but will live much longer than that on average. HDD manufacturers usually rate their HDDs in terms of number of cumulated load/unload cycles, of Annualized Failure Rate (AFR), and Workload Rate Limit (WRL). e.g. https://www.seagate.com/www-content/datasheets/pdfs/skyhawk-3-5-hddDS1902-7-1711GB-en_GB.pdf – PierU Oct 22 '22 at 13:42
  • @PierU: That's quibbling. "Warranty" is the same as "expected healthy lifetime". (But I think we have just about exhausted the subject.) – harrymc Oct 22 '22 at 13:53
  • *"Warranty" is the same as "expected healthy lifetime"*: definitely, absolutely **NOT**. – PierU Oct 22 '22 at 13:55
  • @PierU: Expected and signed-off by the manufacturer. – harrymc Oct 22 '22 at 14:01
  • Still definitely not. If it was, the manufacturer would have to refund half the number of drives he sells. – PierU Oct 22 '22 at 18:45