Create a tiny data file with just a small number of atoms that shows the same issue.
Since you seem to be new to this, it is a bad idea anyway to start with large system simulations. You will make mistakes and with a large system you just have to wait much longer. But if you figure out the procedure with a “toy system”, you will learn how to debug your inputs and find good simulation settings for dependable results much more quickly. Also it is much easier to manually validate your files.
I have just demonstrated to you how to do this with the reply here:
You can provide access to larger files by uploading them to dropbox, onedrive, google drive or similar and then providing a link. But don’t get your hopes up. I will not look at large files or check inputs that need a long time to run. I don’t have the time for that and it is not necessary, anyway (see my comments above).
Please note that a simulation that completes does not automatically mean that it is correct. It just means that you did not have any drastic events happening that lead to invalid data or atoms being propelled across the system too fast.
You can see from the discussion of units units command — LAMMPS documentation
what changes when changing units. For metal vs. real, the length unit is the same, but e.g. energy, time, pressure (and derived properties) are different.
Changing the boundary conditions can make a difference when you have non-zero image flags or bonds that now cross the box but previously were between periodic cells. Only a careful check of the data file can tell if that makes a difference in your specific case.