How to get rid of SMART Reliability notifications?

[Andrii Pavlikovskyi]

I have NVMe disk Samsung PM981a 512 GB (MZVLB512HBJQ)
As far as I understand the manufacturer rated disk endurance is 200 TB and SMART test shows that the disk should be like a FAILED one. But in reality such a disk can handle much more writes, I suppose something near 1000 TB. But Proxmox is sending me every day disk failed notification.


Hetzner's answer about this issue:

Dear Client,

'Critical Warning: 0x04' is caused by "Percentage Used" being above 100%. In its own right, this only indicates that the drive is now out of warranty by the manufacturer. However, as long as 'Available Spare' is greater than 'Available Spare Threshold', you can safely ignore this.

Unfortunately, tools like smartctl will report the disk as failed, so you might need some custom filters for your monitoring.

This topic has been investigated and analyzed with our vendors for a very long time. Unfortunately, it is not possible to disable this warning for our use case. If you insist on it nonetheless, we can offer to replace the SSD for you as a gesture of goodwill.

Thank you very much for your understanding.

So the question is: How to make Proxmox work with "Available Spare Threshold" but not with "Percentage Used" as the last one is useless in case of many NVMe\SSD's?

 

[Andrii Pavlikovskyi]

Any thoughts on this?

 

[leesteken]

Any thoughts on this?

smartd/smartctl is not perfect and the SMART attributes are unfortunately not standardized. Maybe you can get the smartd database updated for that drive by sending this informtion upstream or maybe you locally edit the smartd configuration? The smartctl configuration looks very flexible but I have not tinkered with it myself. I'm sure you can make it stop notifying alltogether, if this particular thing can't be fixed.

 

[marcx28]

Have the same problem. Did you ever get rid of the notification?

 

[_gabriel]

btw, don't forget ZFS (for VMs disks Storage) eat not datacenter SSDs.
Here I use ZFS mirror only for PVE OS, but Lvmthin for VM disks Storage.