Error "Invalid Gurobi license" on HPC using machine-file option.

[zxjroger]

My Julia script calculates some linear programming problem using JuMP+Gurobi in parallel on HPC. When I start submit the slurm job using machine-file, I got the following error:

ERROR: LoadError: On worker 2:
Invalid Gurobi license
error at ./error.jl:33
Env at /home/xz49/.julia/packages/Gurobi/GeYlA/src/grb_env.jl:19
top-level scope at none:0
eval at ./boot.jl:331
#101 at /buildworker/worker/package_linux64/build/usr/share/julia/stdlib/v1.4/Distributed/src/process_messages.jl:290
run_work_thunk at /buildworker/worker/package_linux64/build/usr/share/julia/stdlib/v1.4/Distributed/src/process_messages.jl:79
run_work_thunk at /buildworker/worker/package_linux64/build/usr/share/julia/stdlib/v1.4/Distributed/src/process_messages.jl:88
#94 at ./task.jl:358

...and 19 more exception(s).

Here, I use machine-file so that it will parallel on 2 nodes and 10 CPUs per node. When I do not use machine-file and instead use julia -p 20, i.e., parallel on 20 CPUs on the same node, there is no error.

[odow]

This seems like a question for Gurobi support. What license do you have? It probably doesn't support running over multiple nodes.

[zxjroger]

I am using the Gurobi installed on my university's HPC. I guess there shouldn't be a problem with Gurobi support. I tried to use ClusterManager to parallel on multiple nodes, and it works.

[odow]

Glad you got it sorted.

[zxjroger]

For some reason, I still want to adopt the machine-file approach instead of ClusterManager to parallel on multiple nodes. I asked IT support, they told me "I'm not all that familiar with Julia and Gurobi but my guess is that the GRB_LICENSE_FILE environment variable is not being exported to the worker processes." It's weird that if I parallel across nodes using ClusterManager, it works fine. But it does not work for machine-file.

[odow]

Check if the environment variable is being set in your Julia code? This isn't a Gurobi.jl issue. Broader questions like how to use machine_file are best asked on Discourse: https://discourse.julialang.org

[zxjroger]

Ok, thanks.