exercises.Rmd

---
title: "Exercises"
author: "Jonas Hagenberg"
date: "15.3.2022"
output: html_document
---

```{r setup, include=FALSE}
knitr::opts_chunk$set(echo = TRUE)
```

# Exercise 1
The goal is to start a virtual machine with `singularity` and create a simple
container with `R`, `dplyr` and `tidyr`.

## Get to know your VM
1. Create a folder where you want to store your virtual machine configuration
files. (On windows: use the `git bash`):

```{bash, eval = FALSE}
mkdir vm-singularity && \
  cd vm-singularity
```

2. Create a new VM. For this, you can use an already prepared image with the latest Singularity version for which an image for the VM is available:


```{bash, eval = FALSE}
export VM=sylabs/singularity-3.7-ubuntu-bionic64 && \
  vagrant init $VM
```

3. Start the VM:
```{bash, eval = FALSE}
vagrant up
```

4. Open a shell in the VM (so that you have access):
```{bash, eval = FALSE}
vagrant ssh
```

To exit the shell, type `exit`. To shut down the VM, use `vagrant halt`, to
start it again `vagrant up` and for deletion `vagrant destroy`.

## Build your first container
- create a container definition file that contains `R 4.1.0`, `dplyr` and `tidyr`
- include a test in the definition file where you transform the `mtcars` dataset
into the long format and print the head of the long data.frame
- build the container image

You build the container image with the command:
```{bash, eval = FALSE}
sudo singularity build <container_image_name>.sif <container_definition_name>.def
```

HINT: If you want to move a file into the VM, you can use `scp`:

```{bash, eval = FALSE}
scp -P 2222 /path/to/file/on/host vagrant@127.0.0.1:/home/vagrant
```

The password is `vagrant`.

# Exercise 2
The goal is to run an `R` script within the container on the cluster.

- "download" the container image (the `.sif` file) from your VM to a folder on
your computer. You can adapt the `scp` command from above
- upload the image to a folder in your home directory on the MPIP cluster
(it is also ok to directly upload it into your home directory)
- upload the `container_test_r.R` script to a folder in your home directory
- execute the `container_test_r.R` script in the container in an `sbatch` job

If everything works, you see a `container_testdata.Rds` file in your home directory.

HINT: To specify your home directory, you explicitly need to use `/home/<USERNAME>`,
`~` does not work.

# Exercise 3
*This is a bonus exercise for people who are very fast.*

The goal is to mount a folder from the cluster into the container.

1. Create the following directory on the cluster:

```{bash, eval = FALSE}
mkdir ~/container_exercise_3
```

2. Upload the file `testdata_exercise_3.csv` to the folder `container_exercise_3` on the cluster.

3. Upload the file `exercise_3.R` to a folder in your home directory that is not
`container_exercise_3`.

- find a way to use a `singularity` command (together with `sbatch`) that can
execute `exercise_3.R` without modifying `exercise_3.R`