zlacker

I mean to be fair, had this not been a SSD defect, the probability of four of them dying at the same time ( or in extreme close proximity of a few hours ) is indeed very unlikely. And choosing different SSD vendor would have prevented this happening even if one did have counter overflow. Or in this case the 2nd Server could have been a different Model or vendor.

>>ksec+(OP)
I'm not sure if 4 failed for that reason, or if only 2 failed for that reason and then the attempts to restore them from the RAID array failed for a different reason.

2 failures within a few hours is unlikely enough already though, unless there was a common variable (which there clearly was).