Take a minute to determine if your equipment is working properly prior to making any repairs. It will cost you nothing, can do no harm, and is the first step most people miss when they jump to the conclusion their problem is due to software and send for repair.
The overall failure reported by SMART does not occur until after a manufacturer's threshold has been exceeded, which is usually far too late to be useful. The individual counters that report potential failures start moving considerably earlier.
Take a minute to figure out if your equipment is faulty, before you repair anything. Many times the first thing most people do is guess what is wrong with their computer.
If you're using a Windows system then CrystalDiskInfo would work. If you're using a macOS system then DriveDx works. Or simply run smartctl from a terminal. Just remember to install the SMART reader somewhere else than on the hard disk that you are trying to check, since installers also write files onto disks.
Read the actual numbers from the SMART report (you can ignore the "Good / Caution" label). The label represents the manufacturer's chosen failure threshold. Look at reallocated sector count (the amount of sectors which have been identified as being damaged) and current pending sector count (sectors currently unable to be accessed due to inability to access data in order to relocate).
Reallocated sectors are sectors which were determined by the drive to have failed & have been replaced/retired using spare sectors. A handful over years is typical. But dozens or growing amounts of reallocations indicate progressive physical failure.
Pending sectors represent sectors which could not be read when they needed to be relocated, therefore could not be relocated. Any pending sector count greater than zero on a critical drive should warrant your immediate attention.
It is the rate of change which provides information about what is occurring presently, not merely the value of each reading. If a count changes between two different time readings, then you know that the hard drive is failing at present.
SMART is a very helpful tool but it's not a health certificate. This is where your blind spots lie.
SMART shows how much wear has occurred with regard to flash memory — as a percent usage, number of bytes written, etc... SMART barely shows controller health. The controller is what fails spectacularly. A drive may have 96% of its "life" left and completely die out the next day.
Many external enclosure drives don't pass SMART information from the drive to the host system, therefore the utility doesn't report anything for the drive. That is the bridge, not the drive. Removing the disk to test it would be a bad idea if using a WD unit because they encrypt at the bridge level.
When a head crashes, or when a bearing seizes on your hard drive, it happens instantly. There are no counter values that increase before such an event occurs. SMART assumes gradual wear, and not accidents.
When hardware RAID controllers hide each disk from the OS, it can make SMART information per-drive invisible to the operating system. Most hardware RAID controllers have their own tools and alerting that nobody has configured.
More informative than many of the counters. Free to observe.
Freezes that clear on their own. A machine locks up for twenty or thirty seconds with a solid disk light then continues normally. This is a drive trying over and over again to read a sector it cannot read while the operating system waits. This reads as an old computer and it is a dying disk.
Copies that stall at the same point. A transfer will run fast then stop at a consistent percentage every time it hits the same unreadable region.
Files that open on the second attempt. Your first try fails, or it takes twenty seconds. Your second time trying to access the drive is almost immediate. Eventually, your drive will work and cache the results.
Slowness that arrived suddenly. Slowdown in performance due to software build up is gradual; hardware failure results in a significant decline in a short time frame. A computer or laptop that has been slow for a couple of weeks is most likely experiencing a hardware problem.
And one of them does not involve paying anybody.
Counts healthy, machine misbehaving. Software troubleshooting is worth trying. Run the repairs or reinstall the operating system if you need to and just deal with it as an operating system issue which most likely it is.
Counts climbing. Stop the repair work. Not a disk image, not a clone — get out the file manager and transfer to another physical drive all files and data you want to keep. Most important first. Then stop using the drive entirely.
What not to do in the second case: chkdsk /r, defrag or a deep recovery scan. Each will run for many hours scanning every single track of the entire surface of your failing hard disk drive. That's exactly what a failing drive can't withstand. Many drives die during such a scan, even though the scan is meant to save them. If the drive still reads, you're the best recovery tool available — and you cost nothing.
Install CrystalDiskInfo on Windows or DriveDx on macOS, to a different drive than the one you are checking, and look at these values from the results of a test: reallocated sector count & current pending sector count. Don't care about the overall Good/Caution verdict. It's just a manufacturer defined threshold.
Reallocated sectors = bad sectors the drive found and moved to spare space – damage already handled. Pending sectors = bad sectors the drive currently finds unreadable, but has not yet retired because it cannot recover the contents to move them – the more urgent number.
Not necessarily. That’s the threshold, so a drive can pass it even if its reallocation count continues rising. It tells you very little about SSD controllers. It doesn’t tell you anything about drives behind a USB bridge that does not pass SMART through. And nothing about sudden mechanical failure.
Your hard drive is retrying to read a bad sector on the disk. Reading bad sectors can take time. Your operating system simply waits. It reads as an old computer and it is usually a failing disk.
No. When you use chkdsk with the /r option, it reads every single sector on the disk where the disk is having problems, for hours, and fixes problems by deleting those areas it can't fix. If your disk still reads, you should copy all of your files off first, prior to attempting anything else, because that is the only step that helps.
Honestly? No one knows. Some can operate for months while others can fail in a matter of days. One thing we know for sure is that these counts will never decrease — thus the window of opportunity gets smaller and smaller from now on.