Most bizarre SMART report from brand new drives

[ThumbOne]

I bought a pair of Western Digital Green 4TB HDDs off Aliexpress recently. And formatted both with BTRFS, then went to copy a ton load of data onto the first one (migrating from an older 4TB)) using rsync.

A lot of errors were reported. So I ran a SMART long test, and it passed but had 52 Reallocated Sectors reports and three Pre-fail flags set as illustrated in the attacked image. Aliexpress accepted both of them as returns.

Looking into it Greens someone suggested they were SMR (Shingle Magnetic Recording) and I should aim for CMR (Conventional Magnetic Recording), and Blues being a little pricier offered that, so I sunk some money into getting two Western Digital Blue HDDs.

Given the sour experience with those Greens, and thinking the vendor is maybe shipping seconds with a low price tag I checked the Blues SMART attributes and was surprised. Very surprised.

The same three attributes have the P (Pre-Fail) flag set out of the shop (this time from a bricks and mortar store up the road). All the attribute values, raw and scaled look fine, and I ran long tests and both passed with no rsynced the a drive to near full and no errors (Reallocated_Sector_Ct has a raw value of 0). I attach the full SMART reports for both as text files.

What I'd like to know is what's up here. I have some theories (in the poll).

Attachments

Odd SMART 2 - WDC_WD40EZAX-00C8UB0_WD-WX52DC4J05T5_2025-08-12_0954.txt

SMART report from second WD Blue

(8.44 KiB) Downloaded 15 times

Odd SMART 1 - WDC_WD40EZAX-00C8UB0_WD-WX62DC4FNXHD_2025-08-12_0954.txt

SMART report from first WD Blue

(8.37 KiB) Downloaded 14 times

SMART attributes reported by gsmartcontrol for brand new fault reporting WD Green

[ThumbOne]

Given Western Digital is obsessed with Microsoft, I asked Copilot what it thought - it votes 3.

What the P Flag Means

The P flag in SMART attributes stands for Prefailure Warning. It indicates that the attribute is considered critical for predicting imminent drive failure. If the normalized value drops below the threshold, the drive is deemed at risk.

In your report, the following attributes have the P flag:
ID Attribute Name Value Worst Threshold RAW Value Notes
1 Raw_Read_Error_Rate 100 253 051 0 P flagged, but RAW is 0
3 Spin_Up_Time 100 253 021 0 P flagged, RAW is 0
5 Reallocated_Sector_Ct 200 200 140 0 P flagged, RAW is 0

Why This Might Be a Firmware Quirk

You're seeing the P flag set on attributes that are perfectly healthy—RAW values are zero, and normalized values are well above thresholds. That's not typical behavior. Here's why this might be a WD firmware issue:

WD Drives Often Misuse SMART Flags: WD has a history of setting the P flag on attributes that aren't strictly predictive of failure. For example, Raw_Read_Error_Rate is often flagged even though it's not meaningful in consumer drives (unlike enterprise drives where it's more relevant).

Inconsistent Flagging Across Models: You've seen this on both WD Blue and WD Green drives. That suggests it's not isolated to a single product line, but possibly a shared firmware base or SMART schema.

No Errors Logged: The extended SMART error log and self-test logs are clean. If these attributes were truly at risk, you'd expect some logged events or degradation.

High Initial Values: The normalized values (e.g., 200 for Reallocated_Sector_Ct) are unusually high. WD often uses a 200-based scale instead of the more typical 100-based scale, which can confuse SMART monitoring tools.

[fzabkar]

Copilot is wrong. The Prefail flag merely identifies the attribute type. It does not indicate that there is anything wrong. To determine whether that attribute is failing, you need to examine the raw value and threshold level. In your case the non-zero raw values for the Seek Error Rate and Reallocated Sector Count are indicative of a degraded drive.

[fzabkar]

Aliexpress is a source of fakes. Show us the labels and PCBs of your drives.

[ThumbOne]

Copilot is wrong. The Prefail flag merely identifies the attribute type. It does not indicate that there is anything wrong. To determine whether that attribute is failing, you need to examine the raw value and threshold level. In your case the non-zero raw values for the Seek Error Rate and Reallocated Sector Count are indicative of a degraded drive.

Thanks enormously. With that insight in hand to uncover a blind spot in my thinking, I:

1. Googled and found docs thin on the ground in this rspect, but Wikipedia does agree stating: If one or more attribute have the "prefailure" flag, and the "current value" of such prefailure attribute is smaller than or equal to its "threshold value" (unless the "threshold value" is 0), that will be reported as a "drive failure". In addition, a utility software can send SMART RETURN STATUS command to the ATA drive, it may report three status: "drive OK", "drive warning" or "drive failure".

2. Let CoPilot know it was lying to me and it apologised profusely and confirmed your version and Wikipedia's. Fret not I never trust an AI, which is why a post here, where I had hoped to find an experienced eye.

3. Checked a pile of other drives and indeed they all have some attributes witht he P flag set, different in different models.

So the poll concludes with 4. as the elected answer!

[ThumbOne]

Aliexpress is a source of fakes. Show us the labels and PCBs of your drives.

In the HDD space too? I find that challenging to believe. The SSD space, especially USB pluggable is rife with Chinese manufacturers sticking low grade thumb flash drives in a fancier case and fiddling something to fool Windows in particular into reporting a huge capacity that isn't there. But A HDD has a mass/weight quite distinct, you can feel, sometimes hear the spin up and you need to be geared up particularly well in a factory to even enter the market (it's huge entry bar just to make a HDD that works!). To wit I am inclined to believe that some vendors will happily onsell factory rejects (failing prerelease diagnostics or simply from a compromised run at the factory with questionable quality) but do find it hard they'd be able cost effectively to produce complete fake HDDs.

The HDD market is in steep decline and we're down to three manufacturers globally today by accounts I've read (Seagate, Western Digital and Toshiba). I enjoyed this history:

https://www.youtube.com/watch?v=yt5t84Z7u_I

I am deeply unimpressed with Western Digitals customer support but that may not be better anywhere, it the state of customer support in big industry today. Disappointed as they could have taken my first query to them about these P flags and explained what you have, but no, they took both in chat and then by email down some clearly scripted path that involves me buying Windows, installing it on a computer, installing their diagnostic tools, and sending them the report, because smartctl is not trustworthy it seems and they write:

I would like to inform you that our hard drives are not officially tested or supported on Linux operating systems.

Can you believe that? What is left of the HDD market has got to be heavily dominated by servers of every kind not least NAS systems, webservers and database servers and I'll be monkey's uncle if the clear majority of those are'n't *nix systems. I blows my mind that a HDD manufacturer can say this! And that alone elave sme leaning Seagate and Toshiba into the future.

[fzabkar]

https://www.eevblog.com/forum/chat/moved-trying-to-get-ai-to-admit-that-it-lied-to-me/

As for fake HDDs on AliExpress and Amazon, I have seen numerous cases, some in this forum. Typically the vendor will take an old low capacity drive and replace the label. They may even edit the model number in the firmware, clear the SMART attributes, etc.

[ThumbOne]

Capacity on those cheap WD Greens I returned seemed to spec, I almost filled one with rsync. But I wasn't aware of SMART resets (and don't know how to achieve that and am curious - as it implies the SMART API includes write to firmware registers in the HDD) but it could well explain the Reallocated_Sector_Ct jumping fast on first write (disk being old, and with bad sector count rising then counters all reset and onsold). Labels no longer available for inspection as aliexpress has pretty good responsive returns policy, and sent me a shipping label and covered return postage with that so my outlay was a small box at the post office and the trip there.

[fzabkar]

All those nefarious manipulations are done with undocumented vendor specific commands.