Jupyter in Marvin cluster (with Singularity)
Work with Jupyter on Marvin with a Singularity image.
Some of the advantages of this strategy are:
- Independence: there is no need to ask us for installation or updates. Each user will be able to update the Jupyter version or the X package to the last version, in as many images as they need.
- Collaboration: an image can be stored in a lab shared directory, so all the users of the lab may use the same environment.
- Control: it will be necessary to submit a job to work with Jupyter. This will favor the needs of each project, making possible to change the SBATCH parameters to fit into its necessities. Furthermore, it will be easier to control each job via SLURM, solving the memory issues derived from sharing one node for all sessions.
Clone the repository from bitbucket and follow the instructions.