MX500 - SMART's "pending sectors" (0xC5) goes up, then down

[nageme]

There's an odd behavior people have encountered with Crucial MX500 drives.

Sometimes, or under the triggering conditions, SMART's "pending sectors" (0xC5 / 197) goes up gradually, then down quickly (like, a minute).
There's no increase in "reallocation event count" (0xC4).

In another forum someone suggested "the problem could be the cable", but that makes no sense, as the pending sector count is a completely internal state.

Anyone reached any conclusions on what causes the following, and any practical implications?


Here's the pending sector count over a span of some 2 hours, reading the same 50GB area 3 times, with some idle time after each cycle.
The pending sector count goes up from 0, up to 51 while reading, then down quickly during idle. Then, more read/idle rounds. (The time axis isn't linear.)

The drive I tested was only lightly used, and spent most of its time unpowered.
The files I read during testing were written more than 6 months prior, maybe even a year.

Main drive stats:
500 GB, half empty (but probably not actively TRIMmed because of a USB enclosure)
150 powered hours
1.4 TB written
6 average block-erases (0xAD)
1.65 WAF (1 + 0xF8 / 0xF7)
M3CR043 firmware

I'm theorizing, maybe it's related to flash cell charges that dissipating while unpowered, which leads to sectors requiring more error correction. Then, during idle, the drive rechecks or rewrites.
If the theory is correct, an open question is why do pending sectors still collect after the first test round. Though the peak is lower each time.

FWIW, 0xF8 "Background program page count" did increase while doing the reads, but I think less so in later rounds (didn't check carefully).
It also continued increasing slowly during idle time.

 

[natr0n]

Dont bother with this. If it works without issue dont worry. The smart data is bugged on these.

 

[nageme]

"If it works don't worry" kinda beats the whole idea of SMART, which is to get indications before there's trouble.

I'm not sure it's bugged. It might actually be exposing internal workings more than other drives do, and people aren't used to it.
I've since rewritten the whole drive, and been watching it more. The behavior hasn't reappeared (though didn't test thoroughly, just trying to watch it during normal use).

My current theory is still:
0xC5 can show a very dynamic count of sectors that needed more error correction than usual, as judged during read.
When idle, even briefly, the drive rechecks the flagged sectors to decide if they're really suspect. If not, they're unflagged.

I'm not sure about the repeat reads that still flagged some sectors, though less each time.
It could be the same sectors, which suggests that the drive doesn't always rewrite (maybe variability in read "quality" changes the verdict).
Alternatively, it could be different sectors, and they just happened not to cross the threshold the last time they were read.

The drive's background:
Not new, but hasn't been used much at all.
Data was written many months ago, maybe more than a year.
Spent most time unpowered.
Was just occasionally powered, let's say a few minutes to an hour at a time.

Firmware M3CR04* is, I guess, the 3rd hardware generation. The older ...02* and ...03* are separate branches.
Too bad they don't spell it out explicitly in the model name. Would've been better: MX500, MX501, MX502...

 

[natr0n]

Just backup important stuff and go on using it.

I had a mx 100 series 512 gb sata go bad on me long ago data would just disappear. Happened with/on 3 times/drives.

 

[Vincero]

The assumption there is that the C5 attribute is actually being used for that purpose - we don't actually know what it's measuring.

There have been other threads on other forums about this: https://forum.proxmox.com/threads/mx500-ssd-smart-errors.53779/

Long story short... doesn't seem to be something to be worried about unless remapping/reallocation starts to increase and SSD life left decreases.
I've got one of these drives and can't say I've actually ever noticed this but then I don't use it as a main boot drive.

 

[rtwjunkie]

I have 3 MX-500 drives and have not seen the issue. Admittedly this is anecdotal for me. In fact this is the first time I have heard of it. How widespread are the complaints?

 

[nageme]

@rtwjunkie
If I recall correctly, there were more discussions about the raw value going 0 -> 1 -> 0.
But not only.

Related:

#1227 (Crucial/Micron client SSDs: Don't interpret attribute 197 as Current_Pending_Sector) – smartmontools

www.smartmontools.org

 

[Boombastik]

You can read more there:
-https://forums.tomshardware.com/threads/crucial-mx500-500gb-sata-ssd-remaining-life-decreasing-fast-despite-only-a-few-bytes-being-written-to-it.3571220/page-5
See post 116 with the logs.
To catch someone this problem he must open a program like hard disk sentinel and see the logs because the smart attribute goes very quick from 0 to 1 and the 0, when the ssd do flt rewrites.

Two bugs that connected to each, other.

Also update your firmware to the latest M3CR046 , because it fix a hang problem.

 

[nageme]

@Boombastik

Thanks.
Was more than 1 sector pending here, so didn't seem difficult to catch if you're watching it during continuous reads with no idle time.

Not seeing unusual write activity so far. WAF is about +90%.

I'm holding off with the firmware. 045 was buggy, 046 had a problem reported last month.
As far as I'm concerned there's no rush unless it fixes something major.