Hello,

I have bought one of the above model for my Synology NAS. The drive seem to work and perform correctly (although sometimes it is really loud during random seek, but I know enterprise drives aren't tuned for acoustics). What makes me worry is that High Fly Writes and G-Sense Error Rate raw values continously increase. I have overwritten the whole drive with random data just as a burning test, also ran a long DST with Seatools. I attach a picture with the current SMART values. The drive is mounted in az USB enclosure with two layers of foam for the test and it doesn't feel to vibrate at all.

I have checkec Backblaze's SMART data and it turned out they are using dozens of ST10000NM0086s with High Fly Writes and G-Sense Error Rate raw values measuring thousands, and they doesn't seem to connect. Do you think it's normal?

Any comments would be appreciated

Thanks,

Cs.

ISTM that both those results are undesirable. However, if the drive is detecting these events, then one would expect that it would respond to them appropriately.

For example, I would think that a "High Fly Write" would prompt a rewrite of the affected data. That said, the drive only allows for 100 such events before the High Fly Write attribute hits the 0 threshold. Even so, I don't think that this would trigger a SMART failure when it happens.

In short, I don't really know where to go from here. Sorry.

_________________
A backup a day keeps DR away.

Thanks for Your reply. Take a look at the attached picture. These are drives with about 15k POHs. It is very well seem to be usual for this particular drive. It also doesn't seem to be correlated with the degradation of other values like Read Error Rate or Reallocated Sectors Count...

Source: https://www.backblaze.com/blog/smart-st ... ats-remix/



Fun fact: I've reached out to Seagate support and they replied suggesting me to send the drive back since one of the heads might be scratching a platter

The raw value of the G-Sense Error Rate appears to be a rolling average. Typically what happens is that the number of G-Sense Errors is recorded over a predetermined interval and then a rate is calculated for that interval. The calculation is repeated for the next interval, and so on.

The High Fly Writes attribute, however, appears to report the total High Fly Writes over the lifetime of the drive, so the raw value is a cumulative figure. As you can see, the normalised value bottoms out at 1, which is above the threshold.

As for the other "error rates", the raw values are sector counts, not error counts. Seagate's SMART attributes are counterintuitive.

Seagate's Seek Error Rate, Raw Read Error Rate, and Hardware ECC Recovered SMART attributes:
http://www.users.on.net/~fzabkar/HDD/Seagate_SER_RRER_HEC.html

Seagate SMART Attribute Specification:
http://t1.daumcdn.net/brunch/service/user/axm/file/zRYOdwPu3OMoKYmBOby1fEEQEbU.pdf
http://www.hddoracle.com/download/file.php?id=5129

Normal SATA SMART Attribute Behavior (Seagate):
http://t1.daumcdn.net/brunch/service/user/axm/file/Vw3RJSZllYbDc86ssL6bofiL4r0.pdf
http://www.hddoracle.com/download/file.php?id=5130

_________________
A backup a day keeps DR away.

Hi fzabkar, thanks for your information. I have already read those earlier and respect to Your work and for sharing this.

Let's discuss this further trying to understand Seagate's interpretation. While I'm not an expert as you I'm familiar with SMART and I!m not gonna ask why my Seek Error Rate raw values are increasing, or why Raw Read Error Rate isn't 120

For G-Sense Error Rate, the raw value seems to represent the total count of events and the Value is calculated by 100-(lifetime g-sensor error events/1000). Hence at raw 1000 the Value drops from 100 to 99 and so forth. Also when I installed the drive to an Orico eSata dock for testing, G-Sense Error Rate increased even when the drive was idling. Mounting in a closed external enclosure, the raw values only seem to increase while writing to the drive - but not when doing a random seek (read) test. Maybe the sensor is super sensitive, but still unclear. Finally, if you again look at Backblaze's awesome database, you can see that GSER isn't correlated to Power on Hours:



One should think that High Fly Writes could be caused by excessive vibration, but in this case the drive with the most GSER should have the most HFW, but it isn't the case.

Another finding I made is that Hardware ECC Recovered seems to be calculated differently on these drives, note the raw values are still the same as RRER:



After failing to get a confirmation from Seagate support, I have to think that the 189 and 191 increasing is normal, or at least it's not a failure. Otherwise why would Backbalze run dozens of drives with the given values.

Still would love to hear your thoughts.

I have noticed that earlier generations started to report very low normalised values for the Hardware ECC Recovered attribute, typically around 20. This suggests to me that HDDs are now digging data out of noise. Some time later there was a substantial change in the way that this attribute and the Raw Read Error Rate attribute were reported. As you have observed, the maximum normalised value of RRER no longer appears to be 120, but I don't have any current documentation to confirm this.

As for the G-sense Error Rate and High Fly Writes attributes, if Backblaze isn't bothered by them, then I think that their empirical experience should allay your concerns. I did inform Backblaze about Seagate's counterintuitive attributes, but I doubt that they took any notice. At least I didn't notice any amendments to their blog.

https://www.backblaze.com/blog/hard-drive-smart-stats/

_________________
A backup a day keeps DR away.

Yep, I think that concludes the question as of now. Again thanks for your replies.