HOW-TO: Create a MATLAB Virtual Machine

6 Oct 2018 — tutorial

A steadily-increasing number of students and postdocs in our group find themselves needing to run huge numbers of model simulations, and the path of least resistance — using their local desktop or laptop — simply isn't feasible. Thankfully we have the Nectar Cloud at our disposal! I've been helping several people in our group transition to this platform, and this has prompted me to prepare several step-by-step tutorial for the different needs I've encountered so far. It's my intention to make these all available online, and the first cab off the rank is: How to create a MATLAB VM.

Creating a VM

Create a new NeCTAR VM using one of the large templates (e.g., m3.xlarge, which comprises 16 VCPUs, 30 GB root disk, and 32 GB RAM) and the NeCTAR Debian 10 (Buster) amd64 image.

Give it a descriptive name (e.g., username-matlab-vm) and make sure to select a key pair (under “Access & Security”).

Installing required packages

Connect to the VM, noting that the default user account is debian:

ssh debian@144.6.224.174

Once you’ve logged into the VM, make sure that the system packages are all up to date:

sudo apt update
sudo apt upgrade

Restart the VM so that updated kernel images and other system services are reloaded (this will disconnect you from the VM):

sudo shutdown -r now

Install required packages:

sudo apt install gcc g++ gfortran xfce4

Note that xfce4 pulls in Xorg dependencies; MathWorks do not identify a precise list of required Xorg packages:

Even if you intend to run MATLAB only in non-graphical mode, MATLAB requires some X11 libraries to run. Unfortunately, we are not able to specify the X11 requirements down to the exact set of packages required because different distributions may package the libraries differently. MATLAB will work out of the box on desktop installations of any supported distribution.

You could also install OpenJDK for Java support, but the MATLAB installer bundles its own JVM and it isn’t clear whether it will use an installed JVM instead.

Ensure X11 forwarding is enabled

On the VM, make sure that the file /etc/ssh/sshd_config contains the line X11Forwarding yes and that this line isn't commented-out with a #. You can edit this file by running:

sudo nano /etc/ssh/sshd_config

Once you have made this change, press Ctrl+O to save the changes and the press Ctrl+X to exit the text editor.

NOTE: depending on your local computer's configuration, in order for X11 forwarding to work you may have to specify -YC instead of -XC when connecting to the VM.

Installing MATLAB

Create a MathWorks account, if you don’t already have one. Make sure that you use your institutional email address (e.g., username@unimelb.edu.au).
Download the “Linux (64-bit)” installer from the MathWorks website. This will be a single archive named, e.g., matlab_R2018a_glnxa64.zip for MATLAB R2018a.
Transfer the installer to the virtual machine (important: note the trailing colon after the virtual machine’s IP address):
```
scp matlab_R2018a_glnxa64.zip debian@144.6.224.174:
```

Connect to the VM, making sure that X11 forwarding is enabled, and launch the installer:

ssh -XC debian@144.6.224.174
mkdir matlab-installer
mv matlab_R2018a_glnxa64.zip matlab-installer
cd matlab-installer
unzip matlab_R2018a_glnxa64.zip
sudo su
xauth merge /home/debian/.Xauthority
./install -verbose

The graphical installer should appear on your desktop.

During the installation process, associate your MathWorks account with the University of Melbourne’s site license:
- Staff: 24759-36418-71647-13382-89362
- Students: 18098-22076-81253-90866-03229
Install MATLAB to the ephemeral disk (i.e., somewhere on /mnt, such as /mnt/MATLAB/R2018a).
Install the following toolboxes at a minimum:
- The Parallel Computing Toolbox (provides parpool).
- The Statistics and Machine Learning Toolbox (provides betainv, etc).
- The Deep Learning Toolbox, which was previously called the Neural Network Toolbox (provides combvec).
These toolboxes are sufficient for using the LHS framework and running the pandemic influenza simulation model, but you may require additional toolboxes depending on your project.
Once the installation is completed, add symbolic links so that MATLAB is found in your $PATH:
```
sudo ln -s /mnt/MATLAB/R2018a/bin/matlab /usr/local/bin/matlab
```

Launching MATLAB

To use the GUI:

ssh -XC debian@144.6.224.174
matlab &

To use the text interface:

ssh debian@144.6.224.174
matlab -nodisplay

Note that in order to use the Parallel Computing Toolbox (i.e., parpool and parfor) the Java Virtual Machine (JVM) must be enabled, so do not start MATLAB with the -nojvm command-line option.

Other materials

See the Virtual Wranglers community knowledge base, intended for people managing Nectar virtual machines; in particular, see their Nectar HOW-TOs and their Installing MATLAB guide.