NVME Drives reaching their Rated Endurance - Time to replace ?

chatbox

Hi Everyone

I was checking my dedicated Ryzen 5700G server and checked the NVMe SMART data.

The server has 2x Crucial P3 1TB SSDs in RAID1. Performance generally seems fine:

• RAID1 status healthy ([UU])
• No media errors
• No NVMe controller resets or I/O errors
• NVMe temperatures around 34–35°C
• Website TTFB typically 40–70ms
• Network diagnostics look normal

However, both drives report:

critical_warning: 0x4
percentage_used: 100%

Drive 1:

• CT1000P3SSD8
• Data written: ~255 TB
• Media errors: 0

Drive 2:

• CT1000P3SSD8
• Data written: ~247 TB
• Media errors: 0

My Concerns :

1. How concerned would you be as a host or system admin about Crucial P3 SSDs showing 100% percentage_used in a production web hosting server?

2. Have you seen these drives continue operating reliably for a long time beyond the rated endurance?

3. If it were you, would you proactively replace them now, or simply monitor SMART and keep good backups until actual errors appear ?

4. In a real world scenario Could reaching 100% endurance contribute to occasional performance inconsistencies even when SMART shows no media errors and latency remains low ?

Interested in hearing real-world experiences here . Thank you Everyone !

rpqu

@chatbox said:
Interested in hearing real-world experiences here . Thank you Everyone !

https://old.reddit.com/r/DataHoarder/comments/1rdir2j/358_petabytes_written_to_a_256gb_samsung_nvme_its/

Lex

@rpqu said:

@chatbox said:
Interested in hearing real-world experiences here . Thank you Everyone !

https://old.reddit.com/r/DataHoarder/comments/1rdir2j/358_petabytes_written_to_a_256gb_samsung_nvme_its/

Samsung are workhorses, with Kingston I'd say very second best. But I had plenty of Crucials, SanDisks and Adatas fail on me even before their TBW was 1/10 used lol.

CloudHopper

@chatbox said:
3. If it were you, would you proactively replace them now, or simply monitor SMART and keep good backups until actual errors appear ?

You should be keeping "good" backups anyway. All sorts of things can go wrong, not just disk failures, so if the data matters it should be backed up with the 3-2-1 strategy regardless.

chatbox

@CloudHopper said: You should be keeping "good" backups anyway. All sorts of things can go wrong, not just disk failures, so if the data matters it should be backed up with the 3-2-1 strategy regardless.

I'm taking daily backups since long . However the warning is of concern so thought i check here with experts

RapToN

Personally, I would replace one of the SSDs to prevent them from failing one after another without time to replace one.

MikeA

If there's no issues no reason to replace. I've had SSDs go very long past the rated endurance write/reads no problem. I remember my oldest was some old Samsung SATA Evo drives from like 2013. Things lasted forever.

gbzret4d

How critical would be a downtime if the drive fails?
is it your hardware or rented hardware, if rented write the support that they should replace it

gbzret4d

@MikeA said:
If there's no issues no reason to replace. I've had SSDs go very long past the rated endurance write/reads no problem. I remember my oldest was some old Samsung SATA Evo drives from like 2013. Things lasted forever.

Older drives uses more durable nand storages. Tlc is better than qlc,....

DataRecovery

@Lex said: with Kingston I'd say very second best

Technically Kingston doesn't manufacture SSDs and their crucial components.
So you can get a different device even when you buy the same model (especially after a long while).

@MikeA said: I remember my oldest was some old Samsung SATA Evo drives from like 2013. Things lasted forever.

Samsung 840 Pro or 850 Pro from that era (proper, "fat" MLC) most likely can be left to your grandchildren.