Here is some information that might aid in diagnosing the problem.
Early M4A2s, and other vehicles with the GM 6046 twin diesels, used a dry sump engine lubrication system. Later GM 6046 engines used a wet sump engine lubrication system. When this change was implemented there were coincident changes to the layout in the engine bay and the fuel system.
The early fuel system routed the fuel supply from the tanks to the engines, and the fuel return from the engines back to the fuel tanks, through one valve assembly. This system only allowed fuel to draw from one set of tanks, and ensured that the spill returned back to the tanks from which the fuel was withdrawn.
The late fuel system used 4 separate valves to control the fuel supply and direct the fuel return. In this system there were 2 control handles in the fighting compartment, one at each outboard side of the bulkhead. Each of these control handles actuated a rod which acted on an equalizer bar, which then controlled the movement of 2 rods. The outboard rod actuated the upper valve which controlled return flow to the set of fuel tanks on that side of the vehicle. The inboard rod actuated the lower valve which controlled the fuel supply from the fuel tanks on that side of the vehicle. So when a control handle is turned to the "on" position it should simultaneously open both the fuel supply and fuel return valves on that side of the vehicle. All 4 of the valves were identical, and all of them were connected to their control rods by means of an "overload" device. The presumed function of this "overload" device is to allow excess movement of the control rod once the valve had reached its limit of travel.
In this late system the operator could supply fuel from, and return spill to, both sets of tanks at the same time. If the operator chose to only open one control handle and therefore only feed from one set of tanks, the opposite engine would receive fuel by means of a bypass tube which connected the engine side of each fuel supply valve. In this instance, the return flow from both engines is returned to the same set of tanks from which it was withdrawn.
The various anecdotes in this thread seem to indicate that this system has proven problematic for current operators of these vehicles. It appears that this 4 valve system has many potential points of failure, and limited access makes diagnosing the issue difficult. My assessment of the potential problems are: Is the control handle able to turn sufficiently to provide full throw to the upper control rod? Are all three pivot points on the equalizer bar free to rotate so that the lower control rods are able to provide full throw to the overload devices? Are the overload devices working properly to transmit the necessary throw to the valve pistons? Are the valves actually capable of the full range of movement from open to closed and does the valve actually seal in the closed position?
When I finally found the diagram of this late fuel system, it brought back a roughly 25 year old memory of being tasked to observe and report if both lower control rods were moving when the left control handle was rotated. My recollection was that my father did not like my answer when I tried to describe what I saw. What I attempted to describe at that time was that when he rotated the control handle, the outboard lower control rod moved slightly then stopped, the equalizer bar then started to rotate allowing the inboard lower control rod to continue moving a significantly greater distance. After attempting to describe my observations and evidently not communicating effectively, we changed positions so that I was inside rotating the handle and he was in the engine compartment observing the movement of the control rods. I do not recall the precise outcome, but suspect that the investigation stopped because both rods where observed to move somewhat and therefore they must be opening and closing the valves. Now that I know that all of the valves are the same (and therefore should have the same throw), and there is an overload device between the control rod and the valve(which could account for some movement of the rod, but no actual movement of the valve), and that there was definitely an issue with fuel returning to the left set of tanks, I strongly suspect that the linkages were actually opening and closing the lower fuel supply valve, but the upper fuel return valve was staying closed.
Provided the fuel system has not been modified, diagnosis of a fault has the potential to be confusing. To use the original example of the left tank overflowing, if all fuel tanks are full and both control handles are on, the overflow condition could be caused either by a stuck closed supply valve on the left tanks (could also be a pinched hose, or clogged in-tank filter), or a stuck closed return valve (or pinched tube etc.) on the right tanks. If the left supply valve was stuck closed, the left engine would draw fuel from the right tanks via the by-pass tube (unknown to the operator) but spill fuel into the left tanks thereby causing the overflow. If the right return valve were stuck closed, supply for each engine would come from the tanks on their respective sides, but the return from both engines would be directed solely to the left tank, again causing an overflow. A third condition could exist which causes an overflow on the left tanks. If the left control handle was intentionally in the closed position because the operator wanted to only use fuel from the right tanks, but the left return valve is stuck open, the left engine would be intentionally drawing fuel from the right tanks, but unintentionally spilling fuel into the left tanks.
Certainly, diagnosis of a fault with a modified fuel system will require a thorough inspection to determine the actual flow pattern and potential points of failure in the modified configuration.
This video gives a few decent shots of the blue painted fuel system components in the cut-away M4A2 (76) HVSS at CWM. If you pause it at 2:25 it gives a good view of the left fuel supply valve. The lower nipple is where the supply hose from the lower left fuel tank should enter the valve. The upper horizontal nipple is where the by-pass tube from the same location on the right supply valve should connect. The by-pass tube can be seen in this image electrical-tied to the starter. The hose from the valve to the fuel pump is in place.
https://www.youtube.com/watch?v=Gc3gZinTJT8
Colin