Hard drives – still pretty S.M.A.R.T.

Topic

New Reply

HARDWARE By Ben Myers In my last article, Breaking and entering with Linux: What you see (AskWoody 2021-09-27), I said that there were issues with the
[See the full post at: Hard drives – still pretty S.M.A.R.T.]

9 users thanked author for this post.

AlexEiffel, Lars220, lmacri, Job48, SueW, rc primak, oldfry, Alex5723, abbodi86

Reply | Quote

Replies

Hi Ben,

Very nice article. I too replace hard drives if any of the sector counts is not zero. I also replace hard drives if the Reported Uncorrectable Errors is not zero or if the Head Flying Hours gets to be over four years (approx 8800 hours). Furthermore, I don’t trust any new hard drive until its Power-On Hours and Head Flying Hours exceeds 30 days (approx 700 hours) since hard drives have a tendency to fail within the first 30 days or to begin to fail after four years.

I am not keen on letting any hard drive’s temperature get above 50° Celsius while the hard drive is being defragmented since there is the likelihood for SMART to record High Fly Writes (the drive’s heads are flying too high over the disk surface). This could be caused by the disk drive’s temperature being too warm, or can also be caused in newer helium-filled hard drives in which the drive seals have allowed some helium to escape and also allowed some nitrogen and oxygen from the external atmosphere to seep in.

Another thing which I like to periodically check in SMART is the UltraDMA CRC Error Count. This parameter can suddenly become quite high due to either flaky SATA cable connections, or due to failing onboard I/O controller circuitry on the hard drive’s motherboard. I had the latter occur on a hard drive. The result was occasionally obvious system sluggishness and the sudden loss of all system restore points after rebooting the computer. The sudden loss of all system restore points is too much to explain here. I have also had the former occur as well, due to not using latching SATA data cables for the affected hard drive. After replacing the drive’s SATA data cable, I checked the hard drive’s SMART diagnostics in order to see if the drive’s UltraDMA CRC Error Count is no longer increasing. Since the drive’s Ultra DMA CRC Error Count did not increase after replacing the SATA data cable, I considered the hard drive to still be totally good to use since the issue was with the original SATA data cable and not with the hard drive itself.

Interestingly, Piriform’s Speccy program can correctly and fully report a USB attached hard drive’s SMART data for many types of external hard drives which are attached via USB. It all depends on the capabilities of the hard drive’s USB interface. Please see the attached screen capture which shows Speccy’s detected SMART parameters for a Seagate 2TB external hard drive which was attached to my computer via a USB port.

Again, a very nice article.

Best regards,

–GTP

 

2 users thanked author for this post.

Cybertooth, AlexEiffel

Reply | Quote

Yes, Speccy’s (or other programs) info about USB drives is very much dependent on the interface.  I can only conclude that Seagate has built more S.M.A.R.T. smarts into its drive than the ones I have tested.  Needless to say, interesting info.  Thank you… Ben

Reply | Quote

When I contacted Piriform’s new owners, Avast, about Speccy being so out of date, the response can best be summarized as Meh!

Reply | Quote

More often than not, a drive shows up here from a dead system, so I put it into an elderly Thermaltake BlacX Duet, allowing the connection of one or two SATA drives to a USB port, even USB 1.0.  (NEVER try to copy from one drive to another in a BlacX connected with USB!  You’ve never seen slow.)  Neither ClearDiskInfo nor Speccy report any SMART info for a drive in the BlackX unit, which does not handle UASP (USB Attached SCSI Protocol).  I am due for an upgrade, maybe with a StarTech unit, possibly another, to make life easier.  The BlacX also has an eSATA port, which behaves just like any other SATA, and I sometimes use a system with an eSATA port to clone a drive more quickly than with a USB connection.

Reply | Quote

Awesome article on a neglected topic.  THANKS.

I have used Hard Disk Sentinel Pro for over a decade.  As a 30+ year IT tech, I have a number of desktops and laptops in my household.  I have the Family pack (5 PCs Lifetime) of HD Sentinel Pro and it is installed on every one of them.  Currently that costs $53 US.  I’ve never noticed a performance hit using this software.  For each internal hard drive, it provides a status icon in the system tray and an activity icon (once turned on).  The status icon is color coded, and provides the temperature of each drive (F or C).  In my custom built desktop, I have 1 NVMe drive and 6 spinner drives.  You can set custom thresholds for each drive for temperature (NVMe slots on the motherboard have different temp ranges from spinner drives, for example) – 3 different thresholds – below x is normal, x1-x2 is yellow, and above x3 is red.  You can configure audio alarms also and even set Panic Backup and Shut Down for each temp threshold.  It has far too many customizations to list, and a huge amount of info on the Information tab for each drive.

For non-tech users, it summarizes each drive as a percentage of health.  Anything below 100% gets my attention.  The default configuration is pretty good.  I tweak a few settings, but not many.  I was a beta tester years ago for a major brand hard drive manufacturer.  One drive was a large capacity drive (for the time), and so I continued to use it.  After 10 years HD Sentinel warned me when it started getting wobbly.  I replaced it before it failed because of the warning, and had zero data loss.

On the overview tab for each drive, it estimates remaining lifetime, with the standard being “more than 1000 days” and shows Performance and Health as a percentage – starting at 100%, Power on time in days and hours, total start/stop count.

There is an Enterprise version.  The standard version for 1 PC costs $20US, so the family pack is a really good value IMO.

3 users thanked author for this post.

GoneToPlaid, wavy, oldfry

Reply | Quote

Generally, this is a great article about SMART data and mechanical hard drives. However, there are a few points on which I have had a different experience.

“To date, there remain unresolved technical limitations that prevent seeing SMART data of a drive attached to a system on a USB port.”

That depends on how the USB SATA enclosure implements the UASP communication standard, if it does so at all. That’s partly in the firmware of the enclosure and partly in the USB port itself. My Intel NUC Panther Canyon and my StarTech external SSD encloaure (which has a SATA interface) passes SMART data perfectly well through a USB 3.1 or 3.2 gen2 interface on the computer.

CrystalDiskInfo has no trouble extracting data from my externally mounted Samsung860EVO m.2 SATA III SSD in the StarTech enclosure. The internal Biwin SSD (NVME PCIE-2) on the other hand, does not properly parse SMART Data, and very little info can be extracted from it.

Linux tools do exist which do a better job than Disks at reading SMART data from USB external enclosures.

I would not trust any Microsoft or Apple tool which has anything to do with diagnostics, so dumbing down or disabling SMART in either company’s OS or software would be nothing new to me in this regard.

Since SSDs can fail with no SMART data warnings, I look forward to the next article in this series.

 

-- rc primak

Reply | Quote

Great article.

Get up to speed on router security at RouterSecurity.org and Defensive Computing at DefensiveComputingChecklist.com

Reply | Quote

You can improve the output from CrystalDiskInfo by changing the Raw Values column to show Decimal values via the Function Menu.
Go to Function Menu => Advanced => Raw Values and select 10[DEC]
This setting changes the right-hand column of Hex numbers to a more understandable Decimal listing. Most numbers will now show real-world understandable values – although a few will still be weird (mostly temperatures – but those normally show correctly in the top panel).

Also any suspect items will be flagged using the coloured blobs at the start of each line. The coloured blobs change colour for any “caution” or “bad” flags for each attribute. And the summary “Health Status” is very clearly displayed top left (and is also colour coded). This allows even the least computer savvy user to report to me what the screen is showing without having to understand the bulk of the display.

3 users thanked author for this post.

rc primak, AlexEiffel, Paul T

Reply | Quote

Hi Ben:

Thanks for your excellent article. I’d never heard of Clear Disk Info and like the fact that this utility has a clear description for each S.M.A.R.T. attribute in the interface.

I have a Dell Inspiron 15 5584 purchased in August 2019 with a 256 GB Toshiba KBG40ZNS256G NVMe SSD and find that most hard disk utilities only display limited amounts of S.M.A.R.T. data. For example, I don’t see Current / Worst / Threshold data (i.e., showing percentages from 100 to 0 %) for the S.M.A.R.T. attributes in the main CrystalDiskInfo interface, just Raw Values (shown below with Function | Advanced Feature | Raw Values set to 10[DEC] decimal values). See my 22-Nov-2021 post # 2402902 in Deanna McElveen’s Freeware Spotlight – CrystalDiskInfo for further details about my NVMe disk controller and how certain manufacturers like Dell use proprietary sensor output for measuring certain hardware attributes like fan speed. Also note that CrystalDiskInfo never responded to my 15-Nov-2021 support request to explain why the Current / Worst / Threshold columns are hidden (collapsed) in my interface and only shows blank data when expanded.

Also note that there is no Critical Warnings (True/False) section for my NVMe health log shown in my attached Clear Disk Info screenshot.

That being said, both CrystalDiskInfo and Clear Disk Info are much better than Speccy v1.32 (last updated 20-May-2018), which simply reports “S.M.A.R.T. not supported” for my NVMe. See ZloboMiR’s 19-Feb-2020 thread No SSD NVM Info in CCleaner’s Speccy board for more info.

I’ve also attached a screenshot of my hard drive data from HWiNFO (currently my favourite free utility for hardware diagnostics).  I couldn’t fit all the data for my NVMe SSD on one screen so I scrolled down past most of the General Information (e.g., Host Controller, Drive Model, Serial Number, etc.)
——-
Dell Inspiron 15 5584 * 64-bit Win 10 Pro v21H2 build 19044.1415 * Intel i5-8265U CPU * 256 GB Toshiba KBG40ZNS256G NVMe SSD * CrystalDiskInfo Portable v8.13.3 * Clear Disk Info v2.3.2 * Speccy Portable v1.32.740 * HWiNFO Portable v7.16-4650

1 user thanked author for this post.

Cybertooth

Reply | Quote

This is by far the best description of S.M.A.R.T. technology that I’ve ever seen. Hat’s off to Ben Myers for this, and especially for the tip on Clear Disk Info, which seems to go beyond the bare numbers to provide some useful guidance as to what to do with the data.

A lot of (all right, virtually all of) this S.M.A.R.T. data reminds me of the newfangled baseball statistics that have been cropping up like mushrooms in recent years: WHIP, FIP, OPS+, etc. etc. ad nauseam. That’s all fine and good, but if the reader doesn’t have a sense of what a “good” or a “poor” average looks like in these new categories, then the detailed info is well-nigh useless and serves more to befuddle than to elucidate.

Ditto for S.M.A.R.T. data. For example, when a report says (as in Ben’s illustrations) that a drive’s “read error rate” is Current 200, Worst 200, Threshold 51, does that mean that this drive is, at 200, way past (worse than) the threshold of 51, or what? There is no obvious comparator from which to make a judgment.

Better than an answer to that single question would be a comprehensive layman’s guide to the meaning of these decidedly non-self-explanatory categories. Anybody know of one?

 

Keep Running Windows 7 Safely for Years to Come  
Make Windows 10 look and work like Windows 7

Reply | Quote

Cybertooth wrote:

does that mean that this drive is, at 200, way past (worse than) the threshold of 51, or what?

Each manufacturer uses their own values so you need to rely on the SMART program to do the diagnosis (based on manufacturer data). If the program is happy, you should be too.

cheers, Paul

1 user thanked author for this post.

Cybertooth

Reply | Quote

After installing ClearDiskInfo on my Surface Pro 3, I was shocked to find I only had 2% Lifetime remaining on the SSD!!!!! Yikes.

Reply | Quote

Time to make an image backup to external disk, if you haven’t already.

cheers, Paul

Reply | Quote

I had an older one but will make a new one today!!!!

Reply | Quote

I would want confirmation on that 2% remaining – seems a little suspect seeing as it also reports only 2065 power-on hours. I would want to see what Crystal Disk Info (or any other SSD aware SMART Utility) reports before jumping to scrap the drive. It could just be a mistaken interpretation of the SMART data. How old is this drive – 2065 hours implies it is pretty new.

Reply | Quote

The drive in the Surface is an older mSATA type.  If the Surface is constrained on its main memory, this would lead to heavier writing to the SSD, as Windows busily swaps data and program segments back and forth to the SSD.  Stiil, I agree that barely more than 2000 hours of use should not wear out an SSD.

 

Reply | Quote

It would be worth installing Samsung Magician to test that drive – to be sure, to be sure.

cheers, Paul

Reply | Quote

Thanks for the tip Paul, however the Magician says my SSD is not supported by the software package.

Reply | Quote

CrystalDiskInfo?

Reply | Quote

The drive in the Surface is an OEM drive, so Samsung Magician will show exactly nothing about it.  See my article in today’s (2/7) AskWoody.

Reply | Quote

As a novice at tech stuff, I read Ben’s article and found that my SSD was at 53%. So I put in a new Samsung 870 EVO (1tb) and transferred the contents of my old SSD. But when I go into Speccy and also into the properties tab when you right click on the drive in file manager, it says I still have the old SK hynix SC311 SATA 512GB. Did I somehow skip a step when I replaced it? Seems to work fine. It just has an identity crisis . . .

Reply | Quote

Sometimes you need to refresh / rescan the disks for the new one to be picked up.

cheers, Paul

p.s. can I have your old one, should be good for at least 5 years with my usage.  🙂

Reply | Quote