Strictly speaking, thermodynamic entropy is not function of time, but function of thermodynamic state. 2nd law implies only that when adiabatic system goes from equilibrium state 1 to equilibrium state 2, thermodynamic entropy of state 2 cannot be lower than thermodynamic entropy of state 1.
We can introduce more definition of entropy for non-equilibrium processes that is consistent with the previous one for equilibrium states and then 2nd law implies(with caveats) that such entropy cannot sustainably decrease in time, it can only fluctuate - it can spontaneously decrease by small amount for a small time with small probability.
Measuring time flow by entropy increase is a very inaccurate and pointless thing to do. Mechanical or electronic clocks are much better.