I just sent back 2 brand new FireCuda 2TB drives after they failed the SMART self-test. One had some "Reallocated (bad) Sectors" but both of them had incredible numbers in the "Seek Error Rate" and "command-timeout" attributes. The first drive had 3.7 MILLION seek errors and 4.3 BILLION command-timeouts. Which got me wondering...

I have read that the 8 TB Seagate "archive" drive use some sort of fancy caching/delayed writing scheme which causes certain Linux kernels to believe the drive has dropped out (while the drive is taking a breather to flush its cache to disk). The FireCuda is described as havingI suspect that the bad sector count was real. I wonder if the other counts were due to a conflict between the drive and the kernel (CentOS 7.3 3.10.0-514.6.1.el7.x86_64)

Ken

The bad sectors are probably real, but the error rates are not simple error counters. The seek error rate, for example, has a 32-bit count of the total number of seeks in the lowermost bits and a 16-bit count of seek errors in the upper bits. Do a Google search for "seagate error rate" (without the quotes) and you'll find plenty of confirmation.

1 members found this post helpful.

Thank you rknichols - you are a fount of knowledge.

After receiving the second drive - replacement for the one with bad sectors - and starting the SMART test and seeing the seek and timeout pseudo-error counts heading for the sky I went to the Seagate website to find out how to contact their technical support and ask them about it. As a customer I think I would be entitled to ask why their drives seem to show a lot of errors. It appears that they do not want to speak or otherwise communicate with their customers.

I did not think to search to see if an error was an error. I believed my indications which said I had a bunch of errors. I simply sent the drives back to newegg and ordered a Western Digital drive which will serve my needs.

Thanks again for the info. I read several articles on the Seagate error subject. Interesting. I will keep that in mind if I purchase a Seagate drive in the future. At the moment my servers are filled with WD and Toshiba drives and my workstations and desktops are mostly filled with SSDs (Samsung, Crucial, Sandisk and most recently a MyDigital PCIe M.2).

Ken

Your experience with fireCuda drives may not be all that unusual. We have several 2T fireCuda drives in our cluster. All of them are generating errors upon computer restart after hour-long computations. Usually the errors are "READ FPDMA QUEUED" and "ICRC ABRT". It seems that these drives are not compatible with our systems, whereas we are not having these problems with conventional drives. We are replacing all of the fireCuda drives with conventional ones.

Seagate SSHD's in general have had horrible reliability in my experience. My company had dozens of Optiplex 9020's that had Seagate SSHD's, not a single one survived to the end of the warranty. We replaced EVERY SINGLE DRIVE as they failed. A couple machines even had 2 replacement drives during it's warranty period.