When I do this, I get error 2004 immediately, likely due to the existence of rigid body modes. When I apply no-slip to the nodes on one edge, I get error 1009, noting excessive deformation. I am guessing that the balance between gravity and water pressure is off in my implementation.
Also, I've placed the block in the coordinate system so that it is already balanced, so it shouldn't be moving at all if the gravity force and water pressure balance as expected.
As a test, I tried setting the global load as -1 in z and the face load as 1 on the bottom face, hoping they will balance. The same errors arose. Am I misunderstanding how global loads work?
