Other Serial number as read by dmesg(8)

[amnixed]

I have a Teamgroup CX2 SSD 1 TB:

# dmesg | grep ada0
ada0 at ahcich0 bus 0 scbus0 target 0 lun 0
ada0: <TEAM T2531TB W0704A0> ACS-2 ATA SATA 3.x device
ada0: Serial Number TPBF2311070030501712
ada0: 600.000MB/s transfers (SATA 3.x, UDMA6, PIO 512bytes)
ada0: Command Queueing enabled
ada0: 976762MB (2000409264 512 byte sectors)

However, the label on the SSD shows "S/N 1F2312060140308"

The label also shows TPBF-231107003-... but the last several digits don't match what's reported by
dmesg(8).

I checked the "Serial Number" TPBF2311070030501712 with the Teamgroup web site which could not find it. However, the Teamgroup web site correctly identified the SSD when entering the serial number 1F2312060140308 as shown on the label.

I should mention that smartctl(8) read the same "Serial Number" as dmesg(8) did.

I must be missing something...

 

[ralphbsz]

Well, I don't know what you're missing.

Let's get down to fundamentals. Over the hardware interface (in practice that's SATA, SAS or NVME) you can ask a device what its manufacturer, revision, serial number, and importantly WWN is. Here WWN stands for "world wide name", and is a guaranteed unique (see footnote) serial number of the "thing" (see next footnote). The way to read them on FreeBSD is to use the cam control command, with arguments inquiry for SCSI disks, and identify for SATA disks. This is the unfiltered truth of what the storage device tells the computer its identity is. Here's the output from my boot SSD:

device model          CT500MX500SSD1
firmware revision     M3CR023
serial number         1948E22AFDCD
WWN                   500a0751e22afdcd

Now, there is no law that says that the serial number (or WWN) has to be identical to (or even printed on) the paper label. Yet, I've never seen any device where this is not true (except for the footnotes). If you look at disks from high-quality manufacturers (Seagate, WD), they will have both serial number and WWN printed on the label (sometimes in tiny letters, right underneath a bar code). So having a device where it is untrue is highly surprising. Which brings up a question: Is Teamgroup a high-quality or respectable manufacturer? Could it be that they are a cheap fly-by-night outfit, which has crappy device firmware (giving approximate answers for serial numbers), and which has crappy manufacturing practices (like mixing up the labels that get put on SSDs)? I've never heard of them (and I've been working in storage systems for a quarter century), and the few things written about them on the web does not inspire confidence.

So try "camcontrol identify" and see what you find. Educated guess: It will be the same that dmesg and smartctl report. In which case, that tells you something about the SSD manufacturer.

Footnote 1: Yes, I've seen devices with non-unique WWNs and serial numbers. But that mostly happens with prototype devices, which are not sold on the open market. You can imagine the fun of trying to debug a system that has a dozen disks attached, all of which claim to be "Seagate model XYZ revision PROTO serial number 0". The only time I've seen an actual mistake was not with disk WWNs, but Ethernet port MAC addresses: We bought a few dozen motherboards from a respectable manufacturer (Tyan or Supermicro), and two had identical MAC addresses, both when reading them electronically, and on the printed label on the board.

Footnote 2: SATA disks have one port, one serial number, and one WWN. Dual-ported SAS disks can be a little more complicated: while they have one serial number (which will show on both ports as the same), they typically have a total of 4 WWNs: One for the device itself, one for LUN zero on the device (every hard disk has at least one LU, and on normal disks it will be numbered zero), and one each for the two ports. Typically they use a very simple system to make those predictable, for example if the WWN for the device is X, then LUN zero will have X+1, port A will have X+2, port B X+3, and WWNs X+4 through X+7 are unused, and the next disk from the manufacturing line will be X+8.

 

[%pupil -sceptical]

I must be missing something...

I like to add something to what ralphbsz explained:
Not every manufacturer sticks to norms, especially not if they are more of the kind 'most do it this way' but not being in some legal form like a law.
Especially the way serial nunbers (SN) are given to their products is something very individual every manufacturer deals on its own. If you compare SN for similar devices but different manufacturers you'll see directly that the system of how SN are given significantly differ.

SN are part of a company's process to deal with things internally within their organization (enterprise resource planning, data base, back office,...)
Besides SN, and world wide unique numbers (if there is any; which cannot (shall not) be chosen randomly on its own, like e.g. MAC-adresses, EAN, ISBN, WAN-IP) some manufacturers also have additional numbers because of how they deal with things internally.

However,
crucial point is:
As long as the customer is able to surely assign a device (drive) by the identification the system gives to one printed on the hardware, the essential target is reached.
Which actually is always given, as long as you don't catch some cheap crap by a garbage-out-putter.
But that's most you can expect.

If this is not sufficient for your situation for whatever reason,
e.g. you're dealing with so many drives you read and identify drives automatically,
so you need to ensure the system how and which numbers are given on all your drives is reliably the same,
you might think about to stick to products from one manufacturer only, whose system suits your situation best.
Or you may also contact the manufacturer, and ask about his system (which as we all know especially with large companys may only produce such useful answers like:
"Thank you very much for your interest in our products. Please refer the FAQs first. We're happy to announce you now receive all informations on all of our products and our weekly newsletter.")
(Very) large customers may also get special treatment like additional individual numbers.

Anyway
I recommend to use labels on all drives (I also put stickers with the label written on at my drives)
You need to identify and assign the drive once, then you deal internally with your own 'serial number's system'
Will spare you lots of work, and trouble.

 

[amnixed]

Thank you, very educational.

# camcontrol identify ada0
pass0: 600.000MB/s transfers (SATA 3.x, UDMA6, PIO 512bytes)

protocol              ACS-2 ATA SATA 3.x
device model          TEAM T2531TB
firmware revision     W0704A0
serial number         TPBF2311070030501712
additional product id
cylinders             16383
heads                 16
sectors/track         63
sector size           logical 512, physical 512, offset 0
LBA supported         268435455 sectors
LBA48 supported       2000409264 sectors
PIO supported         PIO4
DMA supported         WDMA2 UDMA6
media RPM             non-rotating
Zoned-Device Commands no
[...]

# camcontrol identify ada0 | grep -i wwn
#

As you can see, no WWN entry in the output nor on the printed label.

I have not had any SSDs from this manufacturer previously. For what it's worth, at Newegg, this particular SSD has better than 4 stars average rating by 1,000+ users.

 

[%pupil -sceptical]

Pardon me to bother you with some experience.

As I said:
"As long as the customer is able to surely assign a device [...] the essential target is reached."
That's what you obviously could.
According to WWN I said

world wide unique numbers (if there is any;

Not every device has one, especially drives don't need such (correct me, if I'm wrong.)
Getting some WWN cost money.
Those numbers need to be managed by some organization. So if a device does not mandatorily needs one, some manfufactures save that cost.
The webpage of your SSD names it as some "beginners/newcomers" model.
It may no 'crap', but it's for sure nothing you expect professional scales.
(Sorry for being educational again.)

If you need to rely on there is one, chose a manufacturer/series/line having those.
But I cannot tell how to evaluate this, except to look at some. (Sometimes you can get a hint from looking into datasheets.)

But

better than 4 stars average rating by 1,000+ users.

is some rough guide you may pick a restaurant, but no real criterion;
especially not on a professional level.
And particulary not in this case, because I bet 99% of those 1,000+ users are not even aware there is a number.
It cost less as another drive with the same capacity, was delivered quickly on time, it does its work, the package was cool designed, had no scratch, and they are simply happy to bought some stuff - five thumbs up.

Sorry for being educational, again.
I will not do this to you anymore.

 

[ralphbsz]

About WWN:

Not every device has one, especially drives don't need such (correct me, if I'm wrong.)

I vaguely remember that every modern SCSI disk that conforms to something like SBC-x (where x is some number) must have a WWN. The way to read them (on the wire) is a bit obscure, it's an Inquiry CDB with an unusual page code. I haven't done that in about 7 or 8 years, so my memory is foggy. But then, there is no legal requirement that disks conform to an up-to-date SBC standard. It's not like George Penokie will send a lightning bolt from his retirement and destroy a non-conforming disk (George was the chair of the SCSI committee for quite a few years).

I don't even know the SATA standards well enough to be able to say whether WWNs are mandatory in SATA or not. And again, who cares, if a vendor doesn't want to conform to the standard, nobody will stop them.

To make matters more interesting, there are several different WWN formats, defined by different organizations, and having different length. The common "5..." one is IEEE and 64 bits long, but I know there are also 48- and 128-bit ones, at least in theory (and in my source code).

Let me be brutally honest: If a manufacturer can't even be bothered to configure their devices to have WWNs, and their serial numbers between internal information and paper label are mismatched, that would ring alarm bells for me. I wonder if their quality control for other matters might also be sloppy. Not that the big manufacturers (Seagate, WD, Toshiba, Samsung, ...) are perfect, but at least they try hard.

 

[%pupil -sceptical]

Frankly I don't know about how disk's WWN are given, (Thanks for educating me! )
but I know it from MAC, IP, EAN, and ISBN.
All those world-wide unique numbers cost money to get one (mostly you get sections instead of a single number; of course you ralphbsz know this.)
So if something does not neccessarily depend on having one, such as drive used within a PC will work without one, the producer may save the cost for it.
Thus may lower the price for private customers who don't depend on that neither.

Obvioulsy the SSD in question does not have any.
If you take a look at its webpage, it's sold as some "beginners/newcomers" model.
It for sure works nicely. But it's simply not made for targeting professional market.

So buying some 'economy' private-end-consumer product which is absolutely alright,
but then measure it at professional scales is missing the point.

 

[PMc]

Footnote 1: Yes, I've seen devices with non-unique WWNs

WWN are not always unique. HP devices have the same:

# (camcontrol identify ada1; camcontrol identify da7 ) | egrep "HP SSD|WWN"
pass11: <HP SSD S700 250GB SN11611> ACS-4 ATA SATA 3.x device
device model          HP SSD S700 250GB
WWN                   50000000000027b6
pass7: <HP SSD S700 500GB SN10033> ACS-4 ATA SATA 3.x device
device model          HP SSD S700 500GB
WWN                   50000000000027b6

Verbatim and SiliconPower don't have any. My other friends seemingly have distinct ones.

Serial Numbers, then, are configured when you load the firmware, as you can do with the software from usbdev.ru. (This is not the firmware upgrade as intended by the manufacturer, but a recovery process in case the device had been bricked)

 

[cracauer@]

I have heard of mismatches between printed and firmware-embedded serial numbers before.

Also note that refurbished drives sometimes have SMART data reset. That can probably affect the serial number.