Academic License for super computing clustersAnswered
I'm working on a project using our university's 3rd generation HYAK cluster (called "KLONE"). I first obtained an academic license under host ID "A". When I ran my MINLP model, I was assigned to a new node with a different hostname and host id "B" so the previous license I requested could not be used. I will be assigned to another host ID "C" if I try to rerun the model. I'm wondering if gurobi has a wildcard for host ID in the license. Or is there another way to get around with it?
We offer free site licenses to academic institutions which are best suited for use on cluster machines. Ask your network administrator at the university to submit a request at support.gurobi.com for an academic site license. You then won't need to worry about licensing for individual machines on the cluster when submitting your jobs.0
I hope you enjoyed a wonderful holiday break. As I'm waiting for my university's decision to obtain an academic license, I have a follow-up question. Is it possible to incorporate the grbgetkey command in my python code to obtain a license after submitting a job to the cluster machines? For example, some lines of code request a license such that the license will only be installed when I run the code on HYAK.
Let me know if I'm explaining myself clearly or not. And I'm happy to provide other information if needed.
You cannot automate the machine licensing process, but an alternative is to use a WLS license. This license is not bound to a specific machine, but it requires that the cluster machines can connect to the internet to contact our license servers, and you must be using Gurobi v10. There are also limits on how many concurrent sessions you can run.
The site license is definitely a better option for computing clusters, but you can try WLS while you are waiting on that to be done. If you run into any issues accessing or using this license, please submit a support request at support.gurobi.com so we can help you directly.0
I ran a python script in which I load gurobipy, but the license type that was loaded is not an academic one despite HRPC staff saying that they do have an academic license. When I create a model, I get: Restricted license - for non-production use only - expires 2024-10-28.
Is creating a python virtual environment and installing gurobipy using pip the reason for this? If so, how could I use the academic license they have while while running a python script on a virtual environment?
When you install gurobipy using pip, there is a trial license automatically included which is restrictive in terms of the size of the models it can solve. However, any other valid Gurobi license on the machine should take precedence. You would need to have a license file at one of the locations listed here: https://support.gurobi.com/hc/en-us/articles/360013417211-Where-do-I-place-the-Gurobi-license-file-gurobi-lic-
Are you running this test on one of the HPC nodes?0
Please sign in to leave a comment.