45

I have a 2TB internal hard drive that might be failing. It spends several minutes at 100% usage when I first turn it on since I installed Windows 10, and yesterday it reported a corrupted Recycle Bin.

I downloaded CrystalDiskInfo, and the pertinent information is reproduced below: enter image description here

In short, the Uncorrectable Sector Count count is 100, and this is a "caution"-level problem on the drive. No other labels are "caution".

I'm not sure what to make of the Current, Worst, and Threshold values. I know that some of their actual values vary by manufacturer, and I've also read that if a current value is below a threshold, that statistic has failed.

The value of the Uncorrectable Sector Count is 100, which is also its worst-recorded value, and the stat has a threshold of zero. Additionally, various other stats read that 100 is both their current and worst while not being marked "caution".

Why is Uncorrectable Sector Count being displayed as caution if various other stats with the same value and threshold aren't? How can a lower-than-threshold value occur with a zero threshold? Is this a sign of imminent drive failure or is the tool merely miscalibrated?

EDIT: Seagate's SeaTools utility has reported that SMART is OK, but it didn't say anything else.

1
  • According to Google's mass statistics, a drive with any uncorrectable sector counts (raw value > 0) are much more likely to fail in the near term HOWEVER this does not guarantee it will fail soon or ever will. As with any situation, make sure you have up to date backups.
    – qasdfdsaq
    Sep 22, 2016 at 13:04

3 Answers 3

17

Why is Uncorrectable Sector Count being displayed as caution?

You need to look at the raw value (8) in this case. There have been 8 events.

The raw value of this attribute indicates the total number of uncorrectable errors when reading/writing a sector.

In this particular case, I would keep an eye on the value and if it continues to increase then replace the drive:

This is a critical parameter. Degradation of this parameter may indicate imminent drive failure. Urgent data backup and hardware replacement is recommended.


S.M.A.R.T. Attribute: Uncorrectable Sector Count

Attribute ID: 198 (0xC6)

Hard drives, supporting this attribute

Samsung, Seagate, IBM (Hitachi), Fujitsu (not all models), Maxtor, Western Digital (not all models)

Description

Uncorrectable Sector Count S.M.A.R.T. parameter is a critical parameter and indicates the quantity of uncorrectable errors. The raw value of this attribute indicates the total number of uncorrectable errors when reading/writing a sector.

Recommendations

This is a critical parameter. Degradation of this parameter may indicate imminent drive failure. Urgent data backup and hardware replacement is recommended.

Source S.M.A.R.T. Attribute: Uncorrectable Sector Count

0
43

To clarify (since this is one of the top responses on Google for this question, but doesn't actually answer it very well, IMHO):

The "Current", "Worst", and "Threshold" values reported by S.M.A.R.T. are normalized qualitative values, that is they don't indicate the actual number of sectors, etc. Instead, they are intended to indicate "how good" or "how bad" that metric's value is on a scale.

With the "Current" and "Worst" numbers, higher is always better, and they are generally normalized to some common scale (such as 0-100). Unfortunately, one of the problems with S.M.A.R.T. is that many of these details are not specified in the standard, so what scale a particular manufacturer uses for these values is completely up to that manufacturer. Usually, most manufacturers seem to use 100 as the top (so the numbers are basically "percent"), but in some cases I've seen 200, or even 253, etc.

So if the scale the manufacturer has chosen is 0-100, then if a metric is reported with a "Current" value of 100 (100% healthy), that means the drive is telling you "It's perfectly OK, nothing to worry about here". If it reports less than that, then it's telling you there may be some cause for concern. What you want to pay attention to is how close these numbers are to the "Threshold" number. If "Current" falls below "Threshold", then that is the point at which the manufacturer considers the drive to be "failed" in some way (and can be sent back for warranty return, etc).

But just because the number hasn't fallen below the threshold doesn't necessarily mean that everything's peachy. It may mean that the drive hasn't failed yet, but it may in the near future. Generally, anything under 100 (particularly on certain metrics which have been shown to be indicators of possible future failure, such as "uncorrectable sector count") could indicate that there's something that you might want to be concerned about.

One of the other problems with these numbers, though, is that there's really no way to tell exactly how the drive firmware is deciding the "quality" of a particular situation. In the example given for the "uncorrectable sector count", the drive is actually reporting a "current" and "worst" value of 100 ("everything's fine!") but if we look at the raw value, it suggests there actually have been 8 uncorrectable sectors encountered. Is 8 sectors a problem? The drive doesn't seem to think so, but it doesn't tell us why. Do you think 8 sectors is a problem? Well, I personally think anything over zero for that metric is concerning, so...

This is why most people will say "just look at the raw value field", but there are two problems with this too: The first is that there's absolutely no standard for what the drive is supposed to return for that value, so you just sort of have to guess what it's reporting there. Quite commonly, for most metrics which involve "counts" of things, the raw value will be the actual raw count (keep in mind that raw values are usually shown in hex), so here it's pretty likely that it's reporting 8 total uncorrectable sectors which have been encountered in this case (which is why CrystalDiskInfo is flagging it as "caution"). The other problem with raw values is that often the drive has a lot more information about the situation that it can take into account than just that one value (for example, exactly when each uncorrectable sector was encountered, whether there have been an increasing number recently, etc), which just isn't reported to us. This may be part of the reason why the drive doesn't seem concerned by the 8 bad sectors (maybe they all happened a long time ago and it hasn't seen any since? But who knows..)

But in general, for a lot of this stuff, you just sorta have to make educated guesses. The 8 in the "raw value" field pretty likely means there have been 8 uncorrected sectors encountered (in the drive's lifetime). The "100" for current/worst says that the drive firmware actually isn't concerned about that and doesn't think it's really a problem ("100% fine"). That's really just a suggestion, though, so your own evaluation may differ.

2
  • 1
    Very clear explanation ! Should be the accepted answer, IMO.
    – ChennyStar
    Oct 31, 2023 at 19:37
  • Very good explanation (in fact, the only proper one I googled. And chatgpt is wrong!) Jan 21 at 15:12
15

As concerns your first question, the three columns together tell you the status of an attribute.

Column "current" means the current value. Usually it is at 100 when everything is ok. Higher values often mean that the attribute has never been updated (implies 100).

The column "worst" tells you what worst value SMART has ever assigned to this attribute.

"threshold" is the absolute health threshold and indicates the value at/below which SMART consideres the attribute a failure. Most attributes that have a zero threshold are not critical. When they decrease, it just means that you drive gets older. Other attributes have thresholds greater than 0 and are often critical.

As concerns your second question, the uncorrectable sector count is an important attribute that very often indicates an imminent total failure. Watch this attribute very closely. If it increments too rapidly, it is better to look for a replacement drive. SMART is not always implemented honestly from what I've seen. 100% health with 8 uncorrectable sectors is such a case. Try to do a surface scan if you would like to save/check the drive, but before you do such an operation with high I/O load, make a backup.

5
  • I did notice that the raw value of the stat reads 0x08 (only after I ask the question, naturally). Perhaps that's why caution is being displayed.
    – Celarix
    Sep 22, 2016 at 11:50
  • Raw values are uninterpreted values from which the health is calculated. These are typically vendor-specific. For example some vendors store temperature in Celsius and others not. They also count sectors/errors/times there and you can't really tell how bad the status is without looking at the current value. Sep 22, 2016 at 11:55
  • No. I do not understand it. RAW value is current value because until now 8 incidents happen. Correct? Threshold value is a bound. When RAW value > Threshold value a warning arises. So, the question remains: If RAW is current value, what is current value and what is worst value? How can current value be 100 when RAW value is 8? Current and RAW are not the same? Please write more.
    – Chameleon
    Jun 9, 2020 at 11:59
  • In this example 8 is the number of sectors affected. Uncorrectable means, well, they cannot be corrected by the logic of the drive. But there are several attributes that can be fixed and they will go back to zero. All thresholds are not just warnings they show clearly that a drive has failed. Raw is the same as current. Current is interpreted by UI and raw is what the attribute really contains. Sometimes the numerical attribute values are flipped (often temperature is shown as quality or real temperature). There is no real standard what a vendor stores as raw value. Jun 9, 2020 at 12:13
  • "The column "worst" tells you what worst value SMART has ever assigned to this attribute.". I don't understand this as "worst" is almost always "100" and another answer suggests that this is the top end of the scale for the healthiness of the metric.
    – HippoDuck
    Jan 30 at 10:01

You must log in to answer this question.

Not the answer you're looking for? Browse other questions tagged .