I run DriveDx in the background. 48 hours ago it started notifying me regularly of read errors on the SSD portion of my Fusion drive. Over an 8 hour period, the read error count got over 2000! About 30 of those were unrecoverable. Right before this started happening, DriveDx reported that the retired block count had increased from 5 to 10.
I immediately made a complete backup of the drive with Carbon Copy Cloner. CCC said it couldn't read one file, and CCC's docs said to delete the file, empty trash, and recover the file from a good backup, which I did. I looked at the SMART status in Disk Utility, which reports "Verified." I ran Apple Hardware Diagnostics, which reported that everything was fine, which seemed odd since the SMART data is generated by the drive itself, right? With DriveDx (and Smart Utility) both reporting over 2000 read errors, it seems that Apple should notice that.
Over the next 24 hours, the read error count rate dropped markedly...only about 125 the next 24 hours. Now, on the 3rd 24 hours, there hasn't been a read error reported for hours.
A bit of history. I bought the iMac Apple Certified Refurbished in early 2017, and it has AppleCare. The wear level on the SSD was at about 97%. Within a month, the retired block count went from 0 to 5. Then a couple of days ago, from 5 to 10. The wear level is down to 86%. I don't understand that. I have a 512GB Samsung EVO SSD that I ran for 2 years as the boot drive in an old MacPro and over that two year period and the wear level went from 100% to 97%. Isn't a 10% wear level decrease in a year high for what I would call "non intensive" use?
So, any advice on the SSD? I can take it in to Apple, but given that their Hardware Diagnostic says no problems and that the read errors seem to have stopped, I figure they'll just say everything is fine.
I immediately made a complete backup of the drive with Carbon Copy Cloner. CCC said it couldn't read one file, and CCC's docs said to delete the file, empty trash, and recover the file from a good backup, which I did. I looked at the SMART status in Disk Utility, which reports "Verified." I ran Apple Hardware Diagnostics, which reported that everything was fine, which seemed odd since the SMART data is generated by the drive itself, right? With DriveDx (and Smart Utility) both reporting over 2000 read errors, it seems that Apple should notice that.
Over the next 24 hours, the read error count rate dropped markedly...only about 125 the next 24 hours. Now, on the 3rd 24 hours, there hasn't been a read error reported for hours.
A bit of history. I bought the iMac Apple Certified Refurbished in early 2017, and it has AppleCare. The wear level on the SSD was at about 97%. Within a month, the retired block count went from 0 to 5. Then a couple of days ago, from 5 to 10. The wear level is down to 86%. I don't understand that. I have a 512GB Samsung EVO SSD that I ran for 2 years as the boot drive in an old MacPro and over that two year period and the wear level went from 100% to 97%. Isn't a 10% wear level decrease in a year high for what I would call "non intensive" use?
So, any advice on the SSD? I can take it in to Apple, but given that their Hardware Diagnostic says no problems and that the read errors seem to have stopped, I figure they'll just say everything is fine.