IBM always had this fight. We actually made it worse when we spec'ed "clockwise at the face of the plug" We meant the "plug", electricians assumed receptacle. The reality was it ended up a 50/50 chance anyway. Everyone had their own way of fixing the pronlem form calling the el;ectrician back to swapping the wires somewhere in the equipment. We were not even consistant where we swapped the wires in our equipment. You could find older boxes that had phases swapping two or three times before it got to the load.
Finally we decided it was going to be clocwise at the face of the receptacle, all the time and we were going to fix our stuff. At the same time engineers put phase detectors and a quick plug swap method of swapping phases in machines that showed reversed phases.
Since we stopped using big motors on stuff I doubt they even care about phase rotation in newer stuff. A "mainframe" these days is a rack full of PC boards and a Disk Array (RAMAC) is a rack full of 3.5" drives.
They don't leak oil on the floor anymore.
