Currently, model repos are zipped, transferred (through scp), and then unzipped every time a job is submitted, regardless of if the repo has changed. This can cause a significant delay even if the model hasn't changed. We need a solution to do this all more efficiently. For example, keeping a zipped copy of each job on each HPC and only updating it as needed (or when specifically requested by an end-user through a route/button).
Currently, model repos are zipped, transferred (through scp), and then unzipped every time a job is submitted, regardless of if the repo has changed. This can cause a significant delay even if the model hasn't changed. We need a solution to do this all more efficiently. For example, keeping a zipped copy of each job on each HPC and only updating it as needed (or when specifically requested by an end-user through a route/button).