New build, disk test results - should I RMA drive(s)?

[demon]

As part of my hardware testing for my new build, I ran jgreco's solnet-array-test-v2.sh on my hopefully-soon-to-be FreeNAS box. It has 6 WD Red 3 TB drives hanging off an LSI SAS 9211-8i host adapter. Its final results were:

Code:

Disk    Bytes Transferred Seconds %ofAvg
------- ----------------- ------- ------
da0         3000592982016   95091     99
da1         3000592982016  110423    115 --SLOW--
da2         3000592982016   98455    103
da3         3000592982016   95757    100
da4         3000592982016  101482    106
da5         3000592982016   73345     77 ++FAST++

Should the two outliers be RMA'd to Western Digital? Or just the too-slow one? Or is this really something I should be concerned about? They are all registered online with WD, so if I have to I can RMA them, but I feel like I should have a more concrete explanation than "well, it's a bit slow..." Thanks for the feedback.

 

[Ericloewe]

da1 isn't that much slower. da5 is surprisingly fast, though (nothing to worry about, of course).

 

[demon]

da1 isn't that much slower. da5 is surprisingly fast, though (nothing to worry about, of course).

Okay, that's what I was wondering. I didn't know if the script's thresholds were actually particularly meaningful, or if the levels were just at the threshold of "oh, that's interesting". Thanks!