Consider the following julia script
#!/path/to/julia --project
using SlurmClusterManager
using Distributed
addprocs(SlurmManager())
i = 0
for i in 1:10
sleep(1)
fetch(@spawnat (workers()[i % length(workers()) + 1]) println(i))
end
When I run this on my local session, it prints every second.
When I run this on the cluster with the slurm manager, it print everything at the termination of the job. This makes it difficult to track the progress of the job.