Train-Test Split & Random_state [duplicate]

I wonder what you need to use random_state in scikit-learn. what number to put for the random_state ? and how do we define the number to the random_state?