Inolvidable

Members
  • Posts

    3
  • Joined

  • Last visited

Converted

  • Gender
    Undisclosed

Inolvidable's Achievements

Newbie

Newbie (1/14)

1

Reputation

  1. Thank you very much Rob. I am also very busy so I totally get it. There is nothing to apologize for. I have been considering the 8TB - 10TB ironwolf drives for a while (great price/performance ratio) but the lack of information about them five or six months after launch seemed odd. I am an "experienced amateur" on HDDs and by sharing these findings I aim to support some information gathering from real experts who can correct and/or complete my post. I hope we can achieve better understanding of the SMART values for these drives so we can sleep better at night after preclearing them.
  2. I have only found 4 SMART reports from either 8TB or 10TB Ironwolfs with enough working hours to get stable normalized values for (1)RRER (7)SER and (195)HER. It is still an small sample to make general assumptions but the data seem to support the hypothesis I posted here that SMART values for the Seagate 8TB drive ST8000DM002 (heavily tested by BackBlaze) could be used as a guide of "normal SMART values" for the new 8TB and 10TB Ironwolf series. The behaviour of Seek Error Rate seems typical for a Seagate drive: A worst value of 60 and a normalized value somewhere between 70 and 85. The Read Raw Error Rate also seem normal compared to the data I have so far. The (195) Hardware_ECC_recovered value is the one that I really want to figure out before I pull the trigger on the ironwolfs. Those values so close to zero worried me. However they seem to be normal for the Seagate ST8000DM002 and for the Ironwolfs I have seen so far.
  3. No, "near threshold" values are almost always false positives, and the feature should probably have been fixed a long time ago. Apart from 2 numbers, it looks like any other modern Seagate SMART report, for first month usage. They have discontinued Runtime_Bad_Block, which doesn't surprise me as it was redundant, confusing therefore. One mildly troubling number is Hardware_ECC_Recovered, which has already dropped to 8. It is encoded into the same number fields as Raw_Read_Error_Rate (same RAW number), but it's not marked as a 'critical attribute' so can't fail the drive. Raw_Read_Error_Rate *is* a critical one, and looks possibly worrisome, having dropped already to 64 (usually 100 or higher for the first years). Everything else looks fine. Since this is the first instance we've seen of the drive, it may be too soon to draw any conclusions yet, about your specific drive. I purchased an Ironwolf 8TB and when my Hardware ECC Recovered (SMART 195) went down to 5 i got really concerned. The HDD also had random loud clicking sounds so it went back to the shop. I could barely find on the web any Ironwolf or any recent Seagate 8TB drive SMART chart for comparison. Thankfully BackBlaze publish the daily SMART status of all their units (along with very cool anual failure rate by brand/model an other stuff) and have just begun the migration to Seagate 8TB drives (ST8000DM002). Backblaze Data: https://www.backblaze.com/b2/hard-drive-test-data.html I imported their logs from September 30 2016 into an excel and customize it so I could see in the same screen smart values 1, 3, 4, 5, 7, 191 and 195 for some of the Seagate 6TB models and 8TB models I know SMART values may change among different models, but I think this chart is interesting: - Smart 1: Raw Read Error Rate Normal Normalized values for the 6TB drives seems to be between 100-120 Normal Normalized values for the 8tB drives seem to be between 70-90 - Smart 7: Seek Error Rate After web research the minimum value for Seagate HDDs seems to be around 60 for most models Normal Normalized values (6TB) between 85-90 Normal Normalized values (8TB) between 80-85 - Smart 195: Hardware ECC Recovered This is kind of interesting: Normalized Values (6TB): 65-75 Normalized Values (8TB): When you see the whole excel with 5000 drives (not just the picture) I can see three groups of values: The percentages are estimated by eye. Not statistical analysis applied Around 1/2 of the 8TB units normalized value of 1 Around 1/4 of the 8TB units normalized value around 12 Around 1/4 of the 8TB units normalized value around 24 If the Ironwolf 8TB or 10TB share some internal design with the ST8000DM002, those low Hardware ECC Recovered values could be normal for this drives What do you think? Edit: Every 8TB unit with Hardware ECC Recovery (HER) = 1 has huuuge numbers (hundreds of millions) in SMART 240 Raw Data (Head Flying Hours) while the ones with HER around 12 or 24 have the Head Flying Hours raw data in the thousands