Another thing I forgot to mention, was another load game exhibited a dynamic light that was stuck on from a long removed lightning bolt.
Its as if some functions are not totally being terminated before a load game.
Something that also may be related is on very odd occasions we get error messages where an entities function is still trying to run after the entity has been ent_removed.
As we get invalid arguments in "Function_name" ENTITY.
We have up to 800 of the same function running at the same time mostly with no errors, but randomly say once a day we may get this problem?
We always check entity pointers before using them and always follow an ent_remove with a return to stop the current function;
I can't help but think all of these may be related.