SLURM: how to run a large job that takes longer than the fair usage policy

I'm not sure if this is the correct stackexchange, so please feel free to point me elsewhere.

I am looking for some advice on SLURM. The HPC cluster I use has a fair usage policy of 48 hours per job, handled via SLURM.

Problem being is that I know my job will take at least 3 or 4 times that, even when requesting max number of cores. I want to avoid splitting up my files (of which the program accepts as input) into many smaller files and start a new job for each file. Does SLURM allow a way to run a job without time limits, but rather SLURM will interactively assign cores/memory to let my program run in a way that is fair to other users. i.e. run on a lower number of cores if other users are requesting.

I basically just want to automate running my program on a file that will inevitably exceed the fair usage policy.

Thanks

Recent Questions...