We have an E3500 that is not under support contract anymore that we
use as a test system for solaris and oracle deployment.

Each line of the internal fc disks (the upper four and the lower four)
is hooked to its own GBIC on a separate i/o board, each upper disk is
mirrored to its lower counterpart.

For a while now I get lots of recoverable read and write errors initiated
by a scsi transport error that affect all disks of the upper row.
No error has been fatal.

I checked the external cabling to the GBICs, I swapped the upper with
the lower disks but the problem is still there.
I ran Sun VTS 6.1 that did not report a problem.

Any ideas what could be wrong?

Uwe