RStudio on Rackham¶
Introduction¶
RStudio is an IDE specialized for the R programming language.
What is an IDE?
See the page on IDEs.
In this session, we show how to use RStudio on Rackham, using Rackham's remote desktop environment.
Forgot how to login to a remote desktop environment?
See the 'Logging in to Rackham' page.
Spoiler: go to https://rackham.uppmax.uu.se/
As RStudio is a resource-heavy program, it must be run on an interactive node.
Forgot how to start an interactive node?
Procedure to start RStudio¶
Below is a step-by-step procedure to start RStudio on Rackham.
Prefer a video?
This procedure is also demonstrated in this YouTube video.
1. Get within SUNET¶
This step is only needed when outside of Sweden.
Forgot how to get within SUNET?
See the 'get inside the university networks' page here
2. Start a Rackham remote desktop environment¶
This can be either:
- Login to the Rackham remote desktop environment using the website
- Login to the Rackham remote desktop environment using a local ThinLinc client
3. Start an interactive session¶
Within the Rackham remote desktop environment, start a terminal. Within that terminal, start an interactive session with 2 cores:
Where:
[naiss_project_id]
is your UPPMAX project code[duration]
is the duration of the interactive session
Resulting in, For example:
Why two cores?
RStudio is a resource-heavy program. Due to this, we recommend using at least two cores for a more pleasant user experience.
What is an interactive node?
Do not start RStudio from the menus
You can start a version of RStudio from the menus. However, this will not have access to loaded modules.
Instead, load RStudio from the module system instead.
4. Load the modules needed¶
In the terminal of the interactive session, do:
How does that look like?
Your output will be similar to:
[sven@r210 sven]$ module load R/4.3.1 R_packages/4.3.1 RStudio/2023.06.2-561
R/4.3.1: Nearly all CRAN and BioConductor packages are installed and available by loading
the module R_packages/4.3.1
R_packages/4.3.1: Note that loading some spatial analysis packages, especially geo-related packages, might
R_packages/4.3.1: require you to load additional modules prior to use. monocle3 is such a package. See
R_packages/4.3.1: 'module help R_packages/4.3.1'
R_packages/4.3.1: The RStudio packages pane is disabled when loading this module, due to RStudio slowdowns
R_packages/4.3.1: because there are >20000 available packages. *All packages are still available.* For
R_packages/4.3.1: more information and instructions to re-enable the packages pane (not recommended) see
R_packages/4.3.1: 'module help R_packages/4.3.1'
RStudio/2023.12.1-402: Sandboxing is not enabled for RStudio at UPPMAX. See 'module help RStudio/2023.12.1-402' for more information
What happens if I do not load R
or R_packages
?
Then you will have the sytem-wide R version 3.6.0 without any packages installed.
What does 'Sandboxing is not enabled for RStudio at UPPMAX' mean?
Nothing.
Here is how it looks like:
[sven@r482 sven]$ module load RStudio/2023.06.2-561
RStudio/2023.06.2-561: Sandboxing is not enabled for RStudio at UPPMAX. See 'module help RStudio/2023.06.2-561' for more information
[sven@r482 sven]$ module help RStudio/2023.06.2-561
--------------------------------------------------- Module Specific Help for "RStudio/2023.06.2-561" ----------------------------------------------------
RStudio - use RStudio 2023.06.2-561
Version 2023.06.2-561
With the Linux distribution used on most UPPMAX clusters (CentOS 7), RStudio/2023.06.2-561
prefers to use a 'suid sandbox'. We do not enable this at UPPMAX. Instead, we disable sandboxing
during startup of RStudio by defining a shell alias for the 'rstudio' command. You may notice
additional errors in the terminal window from which you ran the 'rstudio' command. This is
expected and does not affect RStudio operation.
For performance reasons, UPPMAX disables checks for updates.
UPPMAX also disables the 'Packages' pane of RStudio if an R_packages module is loaded.
5. Start RStudio¶
With the modules loaded, start RStudio from the terminal (on the interactive node):
RStudio can be slow to startup, as R has thousands (!) of packages. Additionally, at startup and if enabled, your saved RStudio workspace (with potentially a lot of data!) is read.
How does RStudio look on Rackham?
RStudio when starting up:
RStudio when started up:
RStudio when ready:
RStudio in action:
The RStudio debugger, at the error message level:
The RStudio debugger, at the function-that-caused-the-error level:
The RStudio debugger, at the program level: