As of writing, my NVMe SSD is about 4 years outdated—sufficiently old to make me take into consideration changing it. That mentioned, with sky-high storage prices proper now, a part of me desires to carry off till the market settles down. The smart transfer in that state of affairs is to verify whether or not your drive actually wants changing—however I didn’t anticipate I’d have to change working techniques simply to seek out the solutions I used to be on the lookout for.
Home windows mentioned my SSD was wholesome
Nevertheless it felt—as the children are calling it nowadays—sus
Home windows gives very primary instruments for monitoring SSD health. Usually, you’d open the terminal and run the next command:
Get-PhysicalDisk | Choose FriendlyName, MediaType, HealthStatus, OperationalStatus, Measurement
Nevertheless, it solely tells you whether or not your storage system is wholesome or not. That’s helpful, but it surely felt a bit too binary for my liking. I needed to know how wholesome it was. It was much less of a qualitative query and extra of a quantitative one.
Sadly, none of Home windows’ built-in instruments had that reply, so I put in CrystalDiskInfo—which, frankly, I ought to have been working years in the past. The distinction in info density was vital. Past the essential “Wholesome” label, it confirmed the drive’s full S.M.A.R.T. knowledge: whole knowledge written, whole knowledge learn, power-on hours, out there spare capability, and—most significantly for my functions—a “Proportion Used” determine that displays endurance consumption based mostly on the producer’s TBW (terabytes written) score.
Why the info felt worrisome
My Essential P3 500GB NVMe drive—the one holding my Home windows set up—presently exhibits a well being score of 77%. On condition that I’ve been utilizing it for roughly 4 years, the tough math suggests I’ve nonetheless received one other 12 months or 12 months and a half earlier than it drops under 70%—a degree the place you must truly begin fascinated with a alternative.
However that assumes SSD put on follows a linear sample. What if it doesn’t? What if degradation accelerates?
A drive doesn’t all the time fail predictably. It will probably report 100% health one day and die the next. That’s why uncooked well being percentages solely inform a part of the story. You additionally want to take a look at the errors accumulating beneath.
And whereas my drive confirmed zero essential warnings, CrystalDiskInfo surfaced one quantity that stored bothering me: the “Variety of Error Info Log Entries” was sitting at 6,605. Nevertheless, it didn’t inform me what these errors truly had been. At that time, I had virtually hit the ceiling of what Home windows-based instruments might inform me.
The picture above exhibits the rely at 6,610. It is because I took the picture after I restarted Home windows, which apparently raised the error rely.
Your ‘100% Health’ SSD score is a trap: The hidden signs of silent data corruption
Your SSD could also be hiding errors, and you will not know till it is too late
Linux confirmed me what these SSD errors truly had been
The place CrystalDiskInfo’s path went chilly
I run a Linux-Windows dual-boot setup, so after hitting a useless finish on Home windows, I booted into Linux to see if I might discover some solutions there.
On Linux, there’s a command-line utility known as nvme-cli constructed particularly for inspecting NVMe drives. It talks to the NVMe protocol instantly, which implies it might pull diagnostic knowledge that generic S.M.A.R.T. instruments both can’t entry or don’t expose clearly.
I began with:
sudo nvme smart-log /dev/nvme0
This pulled the complete S.M.A.R.T. log straight from the drive controller and confirmed what CrystalDiskInfo had proven—however with extra element. Accessible spare was at 100%, media errors had been 0, the proportion used was 23%, and the error log entry rely was 6,605—the identical quantity Home windows confirmed.
However this time, I might truly examine what these errors had been with:
sudo nvme error-log /dev/nvme0
The log returned 16 entries. Entry 0—the one accounting for all 6,605 flagged errors—had a standing area of 0x2002, which decodes to Invalid Field in Command: a reserved code worth or an unsupported worth in an outlined area.
Entries 1 via 15, in the meantime, all confirmed an error rely of 0 and “Profitable Completion.”
Translated into plain English, my SSD was effective. There have been no media errors, no knowledge integrity points, and no indicators of bodily NAND failure. The “errors” weren’t indicators of damage—they had been simply invalid instructions logged sooner or later throughout regular operation.
It’s also possible to take a look at your NVMe SSDs for errors
The self-test Home windows by no means supplied me
The nvme-cli utility can do extra than simply learn well being knowledge—it might additionally run built-in diagnostic exams on the drive itself. So I ran a brief self-test utilizing:
sudo nvme device-self-test /dev/nvme0 -s 1
Then I checked the outcomes with:
sudo nvme self-test-log /dev/nvme0
The take a look at got here again clear: Operation Consequence: 0, which implies the drive handed with out detecting any inside points.
You are testing your PC performance wrong—here is the right way
Artificial benchmarks are effective, however real-world benchmarks are a lot better.
You don’t essentially want to put in Linux to run this software
In case you’re on Home windows and don’t run a dual-boot setup like I do, you possibly can nonetheless use nvme-cli with out putting in Linux.
The best methodology is to create a bootable USB with Ubuntu—or any Linux distro you favor—plug it into your PC, and boot into the reside setting. That provides you a full Linux session working instantly from the USB, with out touching your present Home windows set up. From there, you possibly can set up and use nvme-cli to examine your SSD’s well being, run self-tests, and verify error logs. When you’re accomplished, simply reboot again into Home windows as standard.
Source link
