Prerequisites

Before you start reading the documentation and using GenomeDK, there’s a few things that you should know.

Conventions

You will be jumping a lot between different computers. This includes your own computer, the frontend node, and various compute nodes.

Commands that need to be run on your own computer will look like this:

[local]$ echo hello

Note that the prompt says local in square brackets. When you need to run a command on the frontend node, the prompt will instead say:

[fe1]$ echo hello

Here, fe1 is the name of the frontend node. If you need to run a command on a compute node that you started an interactive job on (more about this later), it will be shown like this:

[s03n11]$ echo hello

In this case the compute node is s03n11, but anything that has the format sXXnYY is a compute node.

Learning to use the shell

When interacting with the cluster you will be using a shell on a Linux/UNIX system. If you are not familiar with these concepts we recommend the Shell novice tutorial from Software Carpentry.

What is a closed zone?

The GenomeDK cluster is divided into zones. There is one open zone and multiple closed zones. Most users only need to use the open zone.

Warning

If you are not part of a project that requires a closed zone, you are in our default, open zone and you can ignore this section completely.

For projects with additional security requirements, GenomeDK provides closed zones. A closed zone is an additional layer of security on top of our existing security solutions. Closed zones were previously called jails.

A closed zone provides:

Isolated networking

Users and compute nodes that are not part of the closed zone cannot access or even see nodes that are in the closed zone.

Data access limitations

Users in the zone cannot directly move data in or out of GenomeDK. Instead, data must be imported and exported. Exporting requires approval from the owner of the closed zone. All exports are logged.

Access through a virtual desktop

Users must access the cluster through a special virtual desktop which disallows copy-paste. This means that users cannot leak data by copy- pasting it out of the closed zone.

No node sharing

In the open zone, jobs from multiple users can run on the same compute node. In a closed zone, only jobs started by users in the same zone can share a compute node.

A closed zone is not tied to a single project. Users will need a separate account/user for each closed zone they need access to. If you think that you need a closed zone for your projects, feel free to get in touch.