Workshop date: 10/17/2019
Home directory: 50 Mb storage
Scratch: 10 TB (90 day limit)
Can purchase storage; requires a PTAO
To get on Rivanna, see:
https://www.rc.virginia.edu/userinfo/rivanna/allocations/
Once on Rivanna, best to submit code to compute nodes. They run whether you are logged in or not.
Two ways to connect to Rivanna:
- Open OnDemand (web interface)
- ssh client
To use Open OnDemand:
- go to https://www.rc.virginia.edu/userinfo/rivanna/login/
- Click Launch Open OnDemand
- Log in with NetBadge Credentials
- Go to Interactive Apps and select FastX Web
- Log in again with NetBadge Credentials
- Select MATE and click Launch Session; select MATE again and select Launch (be sure your browser allows pop-ups for this site)
- this launches a desktop environment in your browser; black box icon opens terminal
If you want to use an ssh client, RC recommends MobaXterm for Windows (free home version)
Probably want to use the same version of R that you have on your latop. To see what version you have on your laptop, enter R.version.string
in the R console.
Use modules to choose R version.
Open terminal in FastX desktop.
Use module spider R
to see what versions are available.
If you want to use R/3.6.1, enter module spider R/3.6.1
to see other modules you need to load.
Enter those modules before R. Order matters. For example:
module load gcc/7.1.0 openmpi/3.1.4 R/3.6.1 rstudio/1.0341
If it works, there will be no echo or confirmation. Also, sometimes a module load can be slow (up to 5 minute wait)
When loading RStudio, enter &
(space &) at the end of the module load command to allow access to the terminal while RStudio runs.
In RStudio, you can load the packages you need. If it asks to create a local library, answer yes.
To see what packages you have installed on your laptop: rownames(installed.packages())
We can run stuff in RStudio, but we really should submit to compute node.
Using Open OnDemand:
- Go to Files, Home Directory
- Click the upload button in the upper right corner
Using FastX Desktop:
- Open FireFox
- Go to UVa Box
- Download files
Can use git clone
if you want to clone a repo to your Home Directory
Can also use Globus. See https://www.rc.virginia.edu/userinfo/globus/
Use Rscript
to submit R script from command line:
Example: Rscript hello.R
Use a SLURM script to request resources to run code. The SLURM script will also contain the request to run R code.
SLURM documentation: https://www.rc.virginia.edu/userinfo/rivanna/slurm/
Example SLURM script:
#!/bin/bash
#SBATCH nodes=1 #total number of nodes for the job
#SBATCH ntasks =1 #how many copies of code to run
#SBATCH time=1 12:00:00 #amount of time for the whole job
#SBATCH partition=standard #the queue/partition to run on
#SBATCH account= myGroupName #the account/allocation to use
module purge
module load gcc R #load modules that my job needs
Rscript myProg.R #command line execution of my job
Can write a SLURM script in RStudio; save as a txt with .slurm
extension
Run a slurm script in the terminal. Example:
-bash 4.1$ sbatch job_script.slurm
To display the status of only your active jobs, type:
squeue -u <your_user_id>
To delete a job from the queue, use the scancel
command with the job ID number at the command line prompt:
-bash 4.1$ scancel 18316
SLURM script needs to be in same directory as R scripts.
Can also use RStudio Server:
- go to Open OnDemand
- Interactive App...RStudio Server
- populate fields (need to populate "allocation" field)
To shut down click Delete
Help: hpc-support@virginia.edu