Manufacturer Recertified Seagate Exos X24 24TB

[rduke]

I have performed a SR Level 5 operation on a Manufacturer Recertified Seagate X24 24TB drive. I plan to use this drive, and a few others of its kind, in a personalize NAS that will store family photos, documents, etc as well as serve as a media server for said family. This is important (to me anyway) when evaluating the current status of these drives and their potential longevity.

Total time to perform this operation was 186 hours (almost 8 days!). Looking at the results:

Make: Seagate Exos X24
Model: ST2400NM000H-3KS103
Firmware: SN05

No (zero) sectors were flagged as Defective, Unrecoverable, or Recovered.
ECC Corrected margin lowered to 25/56
Seek Error margin lowered to 34/55
All other SMART margins were 100%

• Relocated Sectors: 90/90
• Cabling Errors: 200/200
• Uncorrectable: 100/100
• Write Errors: 99/99
• Command Timeout: 100/100
• Pending Sectors: 100/100

Looking at the ECC Corrected and Seek Error margin differences, I have questions when referencing against the SR Study Mode guide:

(#1) — With regard to ECC Corrected the guide states this is worrisome but also if the SMART data shows that no sectors were actually relocated then it is still operating within the threshold of a healthy drive. Therefore, how much concern should there be to see that the drive's ECC Corrected threshold has halved?

(#2) — Again with regard to ECC Corrected, I noticed throughout the 8 days of time that this value fluctuated from 18/56, to 55/56, to anywhere in between. Does this mean that the SMART monitor system in the drive is re-evaluating itself based on the sectors being read/written and is not an overall litmus test for the entire drive? In general, why does it fluctuate up and down like this throughout the process?

(#3) — With regard to Seek Errors, I realize I'm still within the threshold—but, how should I evaluate this value with respect to the ECC Corrected? How should I evaluate it overall?

(#4) — Very simply put, (can I / should I) trust this drive and 5 more of its kind to last me 1, 3, 5... 7+ years in a 24-hour always on NAS setup?

I apologize ahead of time as I used iLO software that mounted the SpinRite image, therefore I do not have a write log. Photos of all seven screens are attached. The first portion of sectors on disk were already scanned before I cancelled the operation and resumed; and are therefore missing in these screenshots.

Thank you for your time, and more importantly for your effort put into this software.

 

[ColbyBouma]

You can safely ignore those particular SMART parameters on Seagate drives. Seagate in their infinite wisdom decided to make those mean something different.

 

[ColbyBouma]

This post explains more: https://forums.grc.com/threads/how-...progress-log-from-a-new-drive.1560/post-11778

 

[rduke]

Eight days of anxiety while watching this scan complete, and it boils down to proprietary hijacking of what otherwise is a standard across all other drive manufacturers?! Dang.

Using your reference calculator to convert the values from my SMART monitor results:

ECC corrected raw data:

0x85 translates to "0 errors in 133 operations."

Seek Errors raw data:

0x044262B5 translates to "0 errors in 71,459,509 operations."

Total Writes raw data (not sure if this one is proprietary):

0x15D3FD translates to "0 errors in 1,430,525 operations."
0X12 translates to "0 errors in 18 operations."

Well that is also good news for me. Thank you for the insight!

 

[peterblaise]

Beautiful, exactly what I do with my super-TB drives - SpinRite 6.1
Level 5 BEFORE putting them into use, BEFORE deciding to send them
back or not.

I also make graphs with free HDDScan and HD Tune ( free and fee-pro ).

All of these helps be a confirmation that I got what I paid for, and later,
when testing the drive for whatever reason, these first tests are a prime
comparison piont so I know what to expect.

Luckily, the Seagate Exos X24 ST2400NM000H-3KS103 SN05 is a
Conventional Magnetic Recording (CMR) drive, not an SMR, so the
rewriting was not 'wasteful' in that all sectors only got written the exact
number of times that SpinRite asked them to be written..

But I test Shingled Magnetic Recording (SMR) drives under SpinRite
6.1 Level 5 nevertheless, they just take even longer, because the drive
may write SMR sections amy number ot times during testing!

I thought a 20 TB took ~4 days, so a 24 TB taking 8 days ... seems like a
lot more slow-end testing.

Your LOG shows:

4:19 pm Jan 2nd, 2026 to 12:29 pm Jan 10th, 2026 = ~8 days - wow.

You are testing AHCI SATA3 ~250,532 MB/s peak, the slower parts
dropping to 135,797 MB/s as far as I can see, as expected.
- yeah, HDDs run ~50% slower on the inside tracks compared to peak
promised performance only available on the opitside tracks.

Lemme look up my records - I'll be back on that point.

- - - - -

So, how do you partition it, and what is it used for?

I ask becajse 16,777,200 MB is the largest partition that NTFS will
compress, and I'm looking for maximum storage with minimum
wasted slack, so I wonder how 24 TB is going to be pressed into
service - do tell!

Thanks.

 

[rduke]

Thanks for the confirmation about the drives and the recommendations, I'll be sure to also test with HDDScan and HD Tune and act accordingly. I chose these drives specifically because they were CMR and not the newer SMR!

Just as important though is you just made me double check that I had the drive plugged-in to the correct port on my test machine. I'm using an HP Microserver Gen8 and two of the ports are SATA III 6Gb/s while the other two are SATA II 3.0 Gb/s. I (un)fortunately was connected correctly to receive the best speed.

IRT your comment about partitioning. My overall plan is to upgrade my current TrueNAS system with six of these Exos X24 24TB drives set up in a RAIDZ2 configuration. The end result will be 96TB of storage with a fault tolerance of (2x) drives. I'm looking for the entire build to get me through the next 5-7 years before ever needing to change anything all the while providing the shared network storage I need along with an entire gamut of media serving power and tinkering speeds.

If you are interested in the entire specs:

Motherboard	ASRock Rack W680D4U-2L2T/G5
Network	(2x) Intel X710 10Gbps (Onboard)
Processor	Intel Core i5-14500
Memory	(4x) 32GB Micron DDR5 ECC UDIMM
HDD	(6x) 24TB Seagate Exos X24 24TB in RAIDZ2
SSD	(2x) 2TB Samsung 850 Pro 2.5" SSD
IPMI via BMC	ASPEED AST2600 (Onboard)
OS	TrueNAS SCALE (via NVMe)

Thank you again.

 

[peterblaise]

Oddly, yes, HP often keeps SATA2 sockets for CD drive booting.

The shared SpinRite LOG reported a SATA3 hard drive connected to a
SATA3 socket, so all was good.

But I'll betcha that if you connect the SATA3 drive to a SATA2 socket,
and run a quick SpinRite 6.1 Benchmark, that will confirm similar
throughput for that drive, even via SATA2 - try it and see.

As far as I can tell, current SATA hard drives are limited by the SATA
standard gear, hence SAS moving on to try to get double connectors
and double speed, and Seagate experiementing with drives designed
with 2 head stacks and 2 connectors.

There's apparently only so much data that can be pulled off a moving
magnet via a floating head.

But, read / write speed in real time only has to be appropriate for the
use case.

For example, I have a couple of 20 TB SATA3 drives, partitioned as
16,777,200 MB compressed plus the remainder ~2 TB as additional
partition also compressed, and the drives are connected via USB2
sockets on 1600-to-2400 MHz 2-core 4 GB ram computers serving
vidio and audio media, and it's always smooth watching and listening
from anywhere on the network, never a pause to wait for more data.

So speed and responsiveness are important for copying in and out
while we wait, but for 'merely' reading file contents in real time, peak
speeds are not so critical.

Purpose fit.

I suppose even SMR would be acceptible, video and audio playback
wise.

I just hate the interminable delays waiting for SMR to run Spinrite
Level 5.

- - - - -

SATA3 is about 14 years old, hitting purchasable system boards
around 2011.

In Moore's law of genreations of doubling performance, or halving
size, every 18 months or so, that's about 9 or 10 generations old.

The largest hard drives were about 3 GB, 4 GB just hitting the market.

10 generations of doubling capacity or halfing size from 3 or 4 GB =
only 1.536 TB to 2.048 TB, so instead, getting to 24 TB is a
comparative miracle - Intel should be so productive.

OK, they indistry has, storage wise, in SSD performance being small,
size-wise, fast, and large, storage-wise.

Imagine a 24 TB NVMe running at 10 GB/s instead of 250 MB/s -
when even Seagate's Exos 2X14, which uses dual-actuator
technology, peaks at 'only' 524 MB/s.

Compare:

Seagate Exos X24	24TB	SATA3 HDD	~$380-$600US	~15.8 to 25 $/TB
SanDisk Ultrastar DC SN655	61TB	NVMe SSD	~24000US	~393 $/TB
Solidigm D5-P5336	122TB	NVMe SSD	~$18,974US	~155 $/TB

It's still a miracle to get such incredible advances in the mechanical
universe.

- - - - -

Nice setup, let us know how it goes.