ADATA NVMe "Media and Data Integrity Errors"

[Enmotus]

Hi! Thanks for this wonderful piece of software. We often recommend it to our customers.

We saw an interesting case where a user was having many IO completion errors with his ADATA NVMe. Though HD Sentinel reports the drive is "PERFECT" and at 95% health, there are 33 million reported "Media and Data Integrity Errors", which is a value we would have expected to influence the health number or otherwise be flagged more prominently.

[hdsentinel]

Thanks for your message and kind words and thanks for increasing attention to this information and value.

Seems so interesting. So if I understand correctly, the particular ADATA NVMe SSD does not work correctly, eg. the data area can't be properly read/written due to IO errors and/or stored data corrupted/damaged?
Not sure if it may be related to the NVMe controller driver too as it may affect the operation the NVMe SSD... Would be nice to know that too, just to be sure.

I'll surely examine the situation with more details and I agree that such issues should be reported to increase attention.

Thanks so much for your attention!

[Enmotus]

So if I understand correctly, the particular ADATA NVMe SSD does not work correctly, eg. the data area can't be properly read/written due to IO errors and/or stored data corrupted/damaged?
Not sure if it may be related to the NVMe controller driver too as it may affect the operation the NVMe SSD... Would be nice to know that too, just to be sure.

Thanks for the response. From what I can tell, the user doesn't see any corrupt data; these seem to be transient communication issues. He's using the Microsoft Standard NVM Express Controller driver, for what it's worth, however, the situation is a bit more complicated as this device is one component of a tiered FuzeDrive virtual disk. The driver guys here tell me that from what they can see in our buffer, these are normal IOs to the tier that are being passed to the NVMe driver via storport, which is returning errors to us.

In one 9-day excerpt of the event log, there are nearly 20,000 disk IO retries. In an earlier 3.5 day sample, there were nearly 50,000. There aren't any NTFS errors during these timeframes. There were thousands of application error events in the 3.5 day sample, but after uninstalling the programs in question, there were only 2 in the newer 9-day sample (which is fairly typical for Windows).

Let me know if there's any more information I can try to gather.

[hdsentinel]

> In one 9-day excerpt of the event log, there are nearly 20,000 disk IO retries.
> In an earlier 3.5 day sample, there were nearly 50,000. There aren't any NTFS errors during
> these timeframes. There were thousands of application error events in the 3.5 day sample,
> but after uninstalling the programs in question, there were only 2 in the newer 9-day
> sample (which is fairly typical for Windows).

Thanks so much, I see.

Sounds so inteteresting. To be honest, I'd expect file system damage/corruption or "just" events related to NTFS file system in this case as they usually appear more common / before application error events when there is a problem with the storage device.

Would be nice to know what is the status provided by the manufacturer specific tool (ADATA SSD Toolbox), how it interprets/displays these "Media and Data Integrity Errors" errors too.

Thanks for increasing attention to this case and the details.

[Enmotus]

Hello again!
I didn't get a response from the customer initially, but I pinged him again today, and he has shared the output from ADATA toolbox. It also does not flag the integrity error count as of any particular importance.

[hdsentinel]

Thanks for the update and the image !

Seems so interesting that the manufacturer-specific tool did not report problem in the health and/or by the diagnostics....
But this may be "normal", as sometimes such tools do not report issues/status correctly (eg. it shows "Lifetime writes = N/A", while it could calculate from the proper attribute in the list).

In the meantime, I did an extensive search and verified some 1000's of NVMe reports about this particular value. On some SSDs (with different brands, not only on ADATA) I saw increased count of "Media and Data Integrity Errors" but much smaller numbers, eg. just 100000 or so, but all SSDs seems working correctly, without any data corruption/data loss.

I agree that the attention should be increased in Hard Disk Sentinel by a lower health value in this case: not (only) for the problems detected but also to indicate degradation upon possible new issues. I'm making the required changes and also keep an eye on possible higher values there too (both on this model and also on other NVMe SSDs too).

Thanks for your attention, help and time!