First, let's deal with terms: the second law of thermodynamics isn't a law! The overall increase in disorder - the tendency of things to 'fall apart' - is just a statistical thing: it's about probabilities, not rules. Given enough time, anything can and will happen.
Think of a sealed box full of air. The air is spread out, filling the box evenly. This is the situation known as 'equilibrium'. The air will tend to stay like this, but not always, just on average. The random movements of the air molecules mean that all throughout the box there are tiny volumes in which air is above average density and there are other volumes in which are is below average density. These fluctuations quickly disappear almost all the time. But not forever. Suppose there was a random fluctuation of density which meant that, overall, one half of the box had slightly more air in than the other half. It's overwhelmingly likely the next few random fluctuations will reverse this situation. But it might not happen. It might be that another fluctuation makes the air even denser in the denser half, and even less dense in the other half. And so on. The probabilities are astronomically crazy, but it's entirely possible that after enough time all the air will have moved to one half of the box. Of course, it's very unlikely to stay there. Quicker than the eye can see it will spread throughout the box again. Probably.
So, the second law of thermodynamics isn't a law at all - it's just a tendency based on statistics. Given enough time and a system full of randomly moving particles anything is possible, and everything possible will actually happen.