Software on the SESYNC server (as of September 2021) is organized using environment modules. Modules allow the user to modify their working environment on the server by loading specific versions of software. SESYNC maintains multiple versions of commonly used software on the server so that users can load the version they want to work with. This is great for reproducibility because you can guarantee that the same version of (for example) R is used every time you run a piece of code. And even if SESYNC IT staff run updates and install a new version of R on the SESYNC server, you can still use the older version if you want to ensure that your code runs exactly the same way every time.
Most of the environment modules on the SESYNC server include a specific version of a single piece of software.
For example, if you load the
R/4.1.1 module, you will be able to run R 4.1.1 in your current environment.
You work with environment modules using the
module command followed by a subcommand. Here are the most commonly used subcommands.
The most important subcommand to know is
module load. This command loads an environment module.
If you don’t specify a version, it will load the default version which is typically the most recent stable version. For example,
module load R
loads the default version of R. After loading the module, you can type
R to open an R session, or
Rscript to run an R script.
You can specify which version of a piece of software you want to load. Entering
module load R/3.6.2
will load R version 3.6.2.
This subcommand shows you which modules are available to load. Simply entering
module avail will show all of them,
or you can search by entering for example
module avail Python to see all available modules with the (case-sensitive!)
string “Python” in their names.
This subcommand shows you which modules you currently have loaded in your environment.
For example if you load the
R module and then enter
module list, you will see this output:
[firstname.lastname@example.org@sshgw00 ~]$ module list Currently Loaded Modulefiles: 1) R/4.1.1
This subcommand lets you switch from one version of a module to another. For instance, if you have R 3.6.2 loaded but want to switch to R 4.1.1, use this syntax:
module switch R/3.6.2 R/4.1.1
It is optional to type out the name of the currently loaded module, but mandatory to type out the name of the one you are switching to.
This subcommand removes a module from your current environment, for example
module unload R
If you are using the RStudio or Jupyter servers to run code, it’s unlikely that you will need to worry too much about modules. On the RStudio server, the R module is loaded by default whenever you start up a session, as well as all commonly used dependencies. On the Jupyter server, Python is run with a virtual environment so it isn’t necessary to load the module. On the Slurm nodes, R is also loaded by default.
Everywhere else on the SESYNC system, you will need to use environment modules.
For example, if you are testing out code on the SSH gateway,
you will need to load modules before running any software. Let’s say you want to create a Python virtual environment from the terminal.
You would need to load the module for your preferred version of Python first. The same goes for Slurm jobs that you submit from the SSH gateway.
If the Slurm job uses software that is managed by environment modules, you’ll need to include
module load in your submit script.
The only exception is that the default version of R is loaded by default on the Slurm nodes.
If you use the rslurm package from the RStudio server, Slurm jobs you submit should work fine without loading the R module.
If you’d like to run code with an older version of R in a Slurm job, add a line at the top of your submit script after the
to load that specific module.