Are my hard drives OK health-wise?

I've just installed various S.M.A.R.T. monitoring utilities onto a couple of laptops, and they are giving conflicting results for both hard drives.

Here's a screenshot of the utilities on my ThinkPad T60, which has a 3 year old hard drive:
http://img7.imageshack.us/img7/505/t60smart.png

As you can see on the top left and bottom right, PassMark DiskCheckup and Acronis Drive Monitor both say that all the SMART values are OK. On the bottom you can see the analysis of the drive from hddstatus.com (launched from the SpeedFan program) - it says all the SMART attributes are good, but notes that the drive has been powered on a lot and is becoming old. Then on the top right is GSmartControl, which is complaining about a bunch of the SMART attributes. If I mouse-over the ones in pink, it says "Warning: The drive has a failing old-age attribute. Usually this indicates a wear-out. You should consider replacing the drive." If I mouse-over the ones in red, it says "ALERT: The drive has a failing pre-fail attribute. Usually this indicates a that the drive will FAIL soon. Please back up immediately!"

So which program should I believe here? Is the drive OK or should I replace it?

Now here's a screenshot of the utilities on my MacBook Pro, which has a 1 year old hard drive:
http://img11.imageshack.us/img11/9241/mbpsmart.png

As you can see on the top left, PassMark DiskCheckup says that all the SMART values are OK. In the middle, GSmartControl has flagged a couple of the Reallocation values, and if I mouse-over each one, it says that "The drive has a non-zero Raw value, but there is no SMART warning yet. This could be an indication of future failures and/or potential data loss in bad sectors". On the bottom you can see the analysis of the drive from hddstatus.com (launched from the SpeedFan program) - it says all the SMART attributes are good, but notes that there are 15 reallocated sectors. Then on the right side is Acronis Drive Monitor, which says the Reallocated Sector Count is a "Fail" and the Reallocated Sector Events is a "Degradation". On the Disk Overview tab it tells me "Reallocated Sectors count S.M.A.R.T. attribute reported bad block on the drive. Increasing number of bad blocks may be an indicator of imminent drive failure. In this case frequent backup and hardware replacement is recommended."

So that drive has had 15 reallocated sectors, which is not necessarily a bad thing. Don't most hard drives have reallocated bad sectors when they are shipped brand new from the factory? I'm not sure I need to replace that drive, as Acronis Drive Monitor is recommending.

Reallocated sectors is always a bad thing. It means the drive is failing. One of the few conclusive results of the Google hard drive study is everyone knows to get rid of hard drives with reallocated sectors because their chance of failing soon is very high. Or at least I thought everyone knew

To me a non zero reallocated sectors count is not always a sign that the drive will die soon however it is definitely a concern. At work I have had drives with a few reallocated sectors run for years at work. To me it becomes a problem if the # of reallocated sectors grows over time pointing to a more serious problem that usually ends in the drive dieing in a few weeks. If it does not grow over time it probably is a localized media defect which can be harmless after the bad sectors are reallocated with new ones from the drives spare sectors pool.

Reallocated sectors is not always bad. It is normal for a newer drive that is high capacity drive to have reallocated sectors, the platters are so high capacity now that the sectors are tiny and go bad easy. I get worried when the count starts to get high or starts going up fast. SMART data does not always show you the whole picture of the drive health.

artbird309 said:

Reallocated sectors is not always bad. It is normal for a newer drive that is high capacity drive to have reallocated sectors, the platters are so high capacity now that the sectors are tiny and go bad easy.

Click to expand...

Wrong; they are always bad. "Reallocated sectors" on "large drives" (or any other drive) that appear during or after the manufacturing process are defined and marked as such by the manufacturer before it ships. Such sectors are never supposed to show up on SMART checking programs run after the drive has been purchased.

delerious said:

Don't most hard drives have reallocated bad sectors when they are shipped brand new from the factory?

Click to expand...

As per the above...with both reliable web info stating as such (and my own extensive experience), _any_ SMART warnings in the "Reallocated Sector Count" or "Raw Read Error Rate" rows is cause for concern.

In terms of what you are using for SMART programs, unfortunate that you are not getting the same/similar info for all of them. GSmartControl (currently v0.8.7) is my usual choice in Windows, with updated Smartctl programs manually put in the folder from Smartmontools (currently v6.0.1). Since two of your three tools are confirming issues, it adds reassurance that there actually is a drive problem. You may even want to increase the odds by trying the totally free/lightweight HDDScan and/or CrystalDiskInfo as well.

So yes, I'd replace the drive. Especially if you see "Reallocated Sector Count" or "Raw Read Error Rate" increasing at all, for example, during daily SMART checks seen over a week or so. Even with a modern drive's ability to reallocate sectors, there is a good chance of data loss (especially when running CHKDSK, defragging, etc.).

One more thing you may want to do with either GSmartControl and/or HDDScan: perform a "short" or "conveyance" SMART self-test with them on the drive. You can get further info (and further reassurance) on the SMART problems. All SMART tests (including the very long "extended" one) are designed to be non-destructive to data on good drives. Its the drive defects that will cause data loss.

Fritingo said:

Wrong; they are always bad. "Reallocated sectors" on "large drives" (or any other drive) that appear during or after the manufacturing process are defined and marked as such by the manufacturer before it ships. Such sectors are never supposed to show up on SMART checking programs run after the drive has been purchased.

Click to expand...

I apologize, I thought the "Reallocated sectors" included the ones the drive found during the normal process and reallocated without data loss.

Yes, reallocated sectors, had data loss, so the drive remapped it.

I have new drives, that had this issue, and the burnin process got them remapped, and they have run for years, with the <10 reallocated sectors. But from what I have seen, when this number increases a few, each day, over several days, it's going be toast, or if it suddently gets a 100+.

On the other hand, I have scsi disks that have worked perfectly upto 2-4k reallocated sectors, maybe it was just the raid card that help there.

Fritingo said:

Wrong; they are always bad. "Reallocated sectors" on "large drives" (or any other drive) that appear during or after the manufacturing process are defined and marked as such by the manufacturer before it ships. Such sectors are never supposed to show up on SMART checking programs run after the drive has been purchased.

As per the above...with both reliable web info stating as such (and my own extensive experience), _any_ SMART warnings in the "Reallocated Sector Count" or "Raw Read Error Rate" rows is cause for concern.

Click to expand...

So if a drive had reallocated sectors before being shipped, they don't show up in the SMART data, and you don't realize that the drive has an increased likelihood of failing. Is that correct?

Even with a modern drive's ability to reallocate sectors, there is a good chance of data loss (especially when running CHKDSK, defragging, etc.).

Click to expand...

I read before that drives will only reallocate sectors on write operations, not read operations, and it's not possible to lose data while reallocating on a write operation.

One more thing you may want to do with either GSmartControl and/or HDDScan: perform a "short" or "conveyance" SMART self-test with them on the drive. You can get further info (and further reassurance) on the SMART problems. All SMART tests (including the very long "extended" one) are designed to be non-destructive to data on good drives. Its the drive defects that will cause data loss.

Click to expand...

OK I just did the short 2 minute SMART self-test on the drive, and it completed without errors. I'll try the long one later.

Using HDD Guardian 0.4.2 on my particular drive (a Samsung 1TB with 3869 Power On Hours), the values it shows for "Reallocated Sector Ct" are: Current 252, Worst 252, Threshold (blank), When Failed (blank), Raw values 0

My interpretation of this is that the factory reallocated sector count was 252, and that there have been no bad sectors since; it is the raw value that I would want to watch, and since that says 0, it has had no bad sectors since it left the factory. Am I correct?

evilsofa said:

Using HDD Guardian 0.4.2 on my particular drive (a Samsung 1TB with 3869 Power On Hours), the values it shows for "Reallocated Sector Ct" are: Current 252, Worst 252, Threshold (blank), When Failed (blank), Raw values 0

My interpretation of this is that the factory reallocated sector count was 252, and that there have been no bad sectors since; it is the raw value that I would want to watch, and since that says 0, it has had no bad sectors since it left the factory. Am I correct?

Click to expand...

Nope.

The normalized value (aka current) is a normalized index for the raw value. Often it is normalized to 100, but it can be normalized to other values, too. The normalized value counts down until it crosses the threshold value, at which point you have a SMART failure. Some normalized values may be capable of going up as well as down, in which case the 'worst' value is the lowest that the normalized value has ever reached.

That is all how SMART is supposed to work, theoretically. In practice, on HDDs I only look at the raw values of a few critical attributes, like reallocated sectors.

Ah, and the wiki article for S.M.A.R.T. tells me that normalized values ranges from 1 to 253, with 1 representing the worst case and 253 representing the best, so my normalized value may have dropped by 1 (but before or after factory?). In theory the actual number of sectors remapped could be in the raw values column, but apparently vendors are not required to do that. Since I have had no issues with the drive and all other values look good, I'm not worried.

delerious said:

So if a drive had reallocated sectors before being shipped, they don't show up in the SMART data, and you don't realize that the drive has an increased likelihood of failing. Is that correct?

Click to expand...

No. It means that bad sectors that show up during the manufacturing process (which is relatively common) and are marked as such by the manufacturer (so they cannot be seen by SMART or other utilities) are nothing to worry about. They have no bearing on the future likelihood of a drive failing; if they did, the drive would not have been shipped.

Any Reallocated Sectors Count of more than a raw value of 0 on an installed drive with a SMART utility is cause for concern. The amount of concern depends on the rate of increase of such sectors (if it does increase) and so on.

delerious said:

I read before that drives will only reallocate sectors on write operations, not read operations, and it's not possible to lose data while reallocating on a write operation.

Click to expand...

What I meant (in more detail) was that sectors are automatically reallocated by modern drives that are seen as defective during all of write, read or verification operations. Wasn't referring to the actual remapping process; meant that heavy usage of a drive with bad sectors increases your chances of running into a data loss problem.

Once sectors are considered bad by the drive and are remapped, getting data back from them--if such data is not able to be read from them properly--can be a major pain. And yes, I've seen drives be unable to read data properly off various bad sectors that are then remapped.

delerious said:

OK I just did the short 2 minute SMART self-test on the drive, and it completed without errors. I'll try the long one later.

Click to expand...

Good to hear. One thing I forgot earlier about GSmartControl. Besides updating the Smartctl utilities it uses, can also run its built-in update-smart-drivedb.exe program before usage (in its folder). The updated database (currently with a date of Nov 20/12) can allow more specific and/or accurate data for your drives.

Well, the idea is this.

When the drive locates a sectore it is unable to correctly read, it marks it as uncorrectable pending sectors. This means you lost the data.

It doesn't remap it yet though, so you can keep attempting to recover it as many times as you like, millions of times, if you wish, but if it doesn't happen in 10 or so tries, it's unlikely to be recovered.

Then when you finally give up on recovering and write over it, it remaps it, and it goes from unrecoverable pending to reallocated sectors.

The fact you have reallocated means the drive lost the data, if you noticed or cared about that dataloss is a different issue.

Is there any way to tell if I lost data? I see one of the SMART attributes is "Uncorrectable Sector Count" and the raw value is 0. Does that mean I haven't lost data?

Also, isn't it possible for the drive to try to write data to a sector that was previously unused and find out it's bad and then reallocate it, which means there was a reallocation without any loss of data?

delete

No that isn't possible.

If a drive writes to a place, it doesn't perform a read check, therefor it doesn't know if that location was bad. It writes it, with the crc data. On read it checks the data it got matchs the crc, if not it attemps to recover it using the ecc data, if it can't, it goes into uncorrectable sector count and on next write it will then go to reallocated sector count.

Yes, data was lost on the disk, cause you have remapped sectors.

Did you actually have data loss on your computer? I don't know, it might have been something that was written over by an update, it might have been a tmp file, that wasn't needed anyways. It could have been corrected due to raid.

Has anyone here had any luck with repair utilities such as what is built into WD's Data Lifeguard or Hitachi's DFT? I have used them on occasion, but only about 1 in 10 has been successful at returning the drive to a functional state. I was not sure they would hold up long-term so I grabbed the information off of the drives and sent them on to the recycling pile (the ones that worked were older drives which were well out of warranty).

On another semi-related note: do you think that the increased failure rates that many of my colleagues and I in the IT world have seemingly experienced are due to the increases in platter density, decrease in manufacturing quality, or a combination of the two? Or is it something completely different? Maybe we're imagining it?

The reason why I ask is that I have had drives all the way back to 40GB IDE drives that worked for YEARS before developing any signs of trouble, but now I seem to average about 6 months to a year on drives before they start showing reallocated sectors, which tend to escalate over a short period of time until the drive stops functioning. Even the better grade drives like Caviar Blacks don't seem to hold up.

I started buying RE3 and RE4 Enterprise drives and that seems to be working out much better so far. Of course I could simply be crazy.

I believe increased failure rates are mostly due to lower prices.

I started buying RE3 and RE4 Enterprise drives and that seems to be working out much better so far. Of course I could simply be crazy.

Click to expand...

Well unless you have 10s of thousands of drives of each type you can not say much statistically..

MrCaffeineX said:

Has anyone here had any luck with repair utilities such as what is built into WD's Data Lifeguard or Hitachi's DFT? I have used them on occasion, but only about 1 in 10 has been successful at returning the drive to a functional state. I was not sure they would hold up long-term so I grabbed the information off of the drives and sent them on to the recycling pile (the ones that worked were older drives which were well out of warranty).

On another semi-related note: do you think that the increased failure rates that many of my colleagues and I in the IT world have seemingly experienced are due to the increases in platter density, decrease in manufacturing quality, or a combination of the two? Or is it something completely different? Maybe we're imagining it?

The reason why I ask is that I have had drives all the way back to 40GB IDE drives that worked for YEARS before developing any signs of trouble, but now I seem to average about 6 months to a year on drives before they start showing reallocated sectors, which tend to escalate over a short period of time until the drive stops functioning. Even the better grade drives like Caviar Blacks don't seem to hold up.

I started buying RE3 and RE4 Enterprise drives and that seems to be working out much better so far. Of course I could simply be crazy.

Click to expand...

I've taken smart data where it's said the specific LBA of the error and zero'd it for a while on a Seagate drive and it's made the drive functional again. It died again after about a year, did the same thing, used dd to zero a specific LBA/range until it read/worked properly, died again in about 6 months to a year. If the information on the drive was important it woulda been trashed but it was irrelevant. There is no fixing this except for replacement.