Well a non-diagnostic solution would be to replace the affected socal card and the controller in the d1000 to which it connects. The diagnostic option would be do one at a time and see which actually causes the problem. If you have capacity you could add another array and see if you can get mirrored across the two arrays, but that will depend on how much data you can transfer before Veritas decides disks are bad becasue of errors.

I should however point out here that there is a less likely cause for this problem, which is a Veritas software corruption. That scenario has happed to me only once in the past 10 years also resulting in disks get marked as failing, in every other case it has been hardware.