Wirebook › Bonus lesson
Is my hard drive getting old? Reading its health chart
Hard drives and SSDs don't just suddenly break without a word. Most of them keep a little health logbook about themselves, called SMART, and if you know how to read it, it can warn you that it's time to copy your photos somewhere safe. Let's take two old friends to the doctor: Grandpa Spinner, a spinning hard drive, and Granny Chip, an SSD made of memory chips. Change what's on their chart and listen to what they tell you.
Patched knees: worn-out spots
A hard drive stores your files as tiny magnetic marks on spinning metal plates, in billions of little spots called sectors. As it gets older, a spot here and there stops holding its marks properly. The drive keeps a box of spare spots for exactly this, so when a spot wears out it quietly swaps in a spare one and moves your data there. That's a reallocated sector: a patch on Grandpa's knee. A handful that never changes is like an old scar and is fairly common. What matters is whether the number keeps going up. New patches every week mean the surface is wearing out, and that's the drive's way of saying "copy your things".
Wobbly spots and unreadable spots
Sometimes the drive tries to read a spot and can't quite make it out. It doesn't give up straight away: it marks the spot as pending (a wobbly spot) and has another squint later, and only swaps in a spare once it knows. Pending spots are more worrying than patched ones, because the file sitting there may already be damaged. Uncorrectable means it tried everything and the data simply couldn't be read back. Either one means: back up today. And if a hard drive starts making a regular clicking or beeping noise, that's the reading arm or the motor struggling. Stop using it for anything else and copy the most precious files off first, one folder at a time, because every extra hour of spinning can make it worse.
A crackly telephone line is not a sick drive
One reading catches lots of people out: CRC errors. They count mistakes that happened on the way along the cable or through the port, like a crackly phone line where the person at the other end is perfectly well. A drive with lots of CRC errors but no patched or wobbly spots usually just needs a different cable, a different port or the plug pushing in properly. The count never goes back down, so after swapping the cable, check whether it's still rising.
Granny Chip: a notebook that wears out
An SSD has no spinning plates and no moving arm, so it never clicks and doesn't mind being bumped. Instead it stores everything in memory chips, which are a bit like a notebook written in pencil: every time a page is rubbed out and written again it wears a little, and each page can only take so many goes. The SSD spreads writing evenly across all its pages so none wears out early, and it reports how much of its promised writing life is used up (Percentage Used, or wear levelling). It also keeps spare pages to swap in for worn ones. When spare pages drop below its own warning line (often 10%), the overall health check finally says FAILED and many SSDs lock themselves to read-only, so you can still copy things off but not write anything new. Most people never get close: a normal laptop SSD writes so little that its writing life lasts many years.
Why "PASSED" doesn't mean "fine"
Most tools show one big overall result first, like "PASSED", "OK" or "Good". That result only flips to FAILED when a reading crosses the limit the drive's maker set, and those limits are often set very late: a hard drive can have dozens of wobbly spots and still say PASSED. So always look at the individual readings, not just the headline. And the other way round: plenty of drives fail with a perfectly clean chart and no warning at all. A good check-up is reassuring, but it never replaces a backup. The best plan is the boring one: keep a copy of everything that matters on a second drive or in the cloud, so it doesn't matter when Grandpa finally retires.
How to see your own drive's chart
- Windows: the free CrystalDiskInfo shows every reading with a colour (Good, Caution, Bad). In PowerShell,
Get-PhysicalDisk | Get-StorageReliabilityCounter | Format-Listshows wear, temperature and read errors. - Mac: Disk Utility only shows the overall result ("Verified" or "Failing") for internal drives. For the full chart, the free command-line tool smartmontools (
smartctl -a disk0) works, and DriveDx is a friendlier app. - Linux:
sudo smartctl -a /dev/sda(or/dev/nvme0for an NVMe SSD), or the Disks app's SMART Data window. - USB drives: many USB cases and sticks hide the logbook. For a USB hard drive,
smartctl -d sat -aoften gets through; cheap USB sticks usually don't keep one at all. - Got the chart as text? Paste it into the SMART report explainer on Tooltern for the same kind of verdict on your real drive (it runs in your browser, nothing is uploaded).
Related lessons: where deleted files really go · why copying to a USB stick is so slow · how a game remembers where you got to.
Grown-up twin: Tooltern's help pages on an external hard drive clicking or beeping, "You need to format the disk before you can use it" and wiping a drive before giving it away, plus the glossary entry on SMART.