These are chat archives for dereneaton/ipyrad
Yeah @tommydevitt, you can submit the
slurm-jupyter.sbatch script just one time and it will start a server that continues to run for as long as the job wall-time limit will allow. You can connect, disconnect, and reconnect to the juptyer server as many times as you want while it is running. The local
ssh command is used to connect to the remote server, and you can close the ssh connection and reopen it as much as you like without it affecting the jupyter server.
It is not totally necessary to set a 'profile' for the ipcluster instance. Profiles are nice because you can modify them to always use a specific setting, which simplifies things so you don't have to write such a long command for the
ipcluster arguments. But using the default (i.e., no profile argument) is fine. When you call a
.run() argument in ipyrad it does one of three possible things to find an running ipcluster instance: (1) it looks for an instance running on the "default" profile; (2) it looks for an instance running with details specified in the
data._ipcluster dictionary; or (3) it expects you to have passed in a connected ipcluster instance explicitly as an argument.