Question about read_restart and velocity

Hello to everyone.
I want to equilibrate an epoxy glass system at 300K (in which the system is solid) in the NPT ensemble for 50ns.
My system is quite big (16k atoms) and I run my simulation on a cluster which has a limitation on the runs you can perform, only allowing you to run simulations that last 48h . 25ns on this machine will do like 40h so to equilibrate in 50ns I should run two simulations 25ns each.

In the first simulation I use a velocity command to initialize the velocities and when the 25ns are completed I use write_resrart command to save my system.

My question is in the second run for the remaining 25ns should I use velocity command after read_restart command for loading my system or if I use it it will spoil the Dynamics of the system?

Thank you in advance

This is a question you have to answer yourself.

How bad would it be to your system when you would suddenly redistribute all the velocities?
In particular when you are still in the process of approaching equilibrium.

There is no question that resetting the velocities during a production simulation would be a bad idea.

Ok ok , and a final question does the write_restart also saves the velocities besides the positions of the atoms?