Logging on to the cluster

Typically when working in bioinformatics with large datasets we would log into a computing cluster such as the one we have prepared for this course. Outside the course, you would work on the cluster/server of your home institution and log in in a very similar way as we show below. You likely would not need a keyfile for your home institution, but would have a password instead. How do we log in for the course?

Logging on

If you are using a Mac or Linux machine, you will need to open a terminal window. If you are using a Windows machine, start MobaXterm or “Ubuntu on Windows” to get a terminal. For MobaXterm users, we recommend to go to “Settings” and change the “Persistent home directory” to a directory of your choice.

To connect to the Amazon cloud, we use the command ssh.

ssh stands for secure shell and is a way of interacting with remote servers. You will need to log in to the cluster using a keyfile that has been generated for you and sent via email.

To log on, we will need the keyfile you got from Physalia. If this was the cluster/server of your institution, you would not need that but use a password instead.

First, copy the keyfile into your home directory (change the number (123) to the user number assigned to you).

cp c123.pem ~

If you get an error about permissions, try the following:

chmod go= c123.pem
chmod -R u+x c123.pem

Then you should be able to log in with ssh whatever your working directory is. You need to provide ssh with the path to your key file, which you can do with the -i flag. This basically points to your identity file or keyfile (here shown for user 123). For example:

ssh -i "~/c123.pem" user123@54.245.175.86

Of course you will need to change the log in credentials shown here (i.e. the username and keyfile name) with your own. Also be aware that the cluster IP address will change everyday. We will update you on this each day.

The first time logging in you will be prompted to accept an RSA key - just type yes!

Downloading and uploading files

Occassionally, we will need to shift files between the cluster and our local machines. To do this, we can use a command utility called scp or secure copy. It works in a similar way to ssh. Let’s try making a dummyfile in our local home directory and then uploading it to our home directory on the cluster.

# make a file
touch test_file
# upload a file from your computer to the cluster
scp -i "~/c123.pem" test_file user123@54.245.175.86:~/

Just to break this down a little we are simply copying a file, test_file in this case to the cluster. After the : symbol, we are specifying where on the cluster we are placing the file, here we use ~/ to specify the home directory.

Copying files back on to our local machine is just as straightforward. You can do that like so:

# download a file from the cluster to your computer
scp -i "~/c123.pem" user123@54.245.175.86:~/test_file ./

Where here all we did was use scp with the cluster address first and the location (our working directory) second - i.e. ./

Making life a bit easier

If you are logging in and copying from a cluster regularly, it is sometimes good to use an ssh alias. Because the cluster IP address changes everyday, we will not be using these during the course. However, if you would like some information on how to set them up, see here