38 Unix Miscellaneous

38.1 Unix Miscellaneous

Here’s a recording of this lecture (19 minutes 57 seconds):

38.2 Unix Basics slides

38.3 Questions

38.3.1 Question 1

What command would make a symbolic link to /really/long/path/to/data/folder/ and name it data_shortcut? What command would remove the link?

ln -s /really/long/path/to/data/folder/ data_shortcut ; rm data_shortcut ln -s /really/long/path/to/data/folder/ data_shortcut ; unlink data_shortcut ln -s data_shortcut /really/long/path/to/data/folder/ ; rm data_shortcut ln -s data_shortcut /really/long/path/to/data/folder/ data_shortcut ; unlink data_shortcut

38.3.2 Question 2

Which of the following are needed/advised when running an R script on htc? Multiple answers may be correct.

Loading the modules you need (a version of gcc and a version of R) Running the script with a command like Rscript –vanilla my_script.R Including options(echo=TRUE) at the top of the R script so that each line of R code is echoed before its output Doing args <- (commandArgs(T)) and then for (i in 1:length(args)) { eval(parse(text =args[[i]])) } in the script to parse command-line arguments

38.3.3 Question 3

Which of the following are true of Slurm job arrays?

They are an alternative to using a loop to repeat similar tasks. They implement parallelism by running each task in the array as a separate Slurm job. They are slightly simpler to use when the tasks/files you need to loop over are labelled 1, 2, 3, … n. Add #SBATCH –array=1-n to your Slurm script in order implement the array (replacing n with the number of tasks in the array). Inside the script, the variable $SLURM_ARRAY_TASK_ID functions like a looping index variable that you can use to control which task gets executed in each job in the array. However, they can still loop over any list of files by using syntax like that shown in the last slide.