Example NorNet / MONROE experiment repository

This is an example repository with all the scripts needed to manage the whole lifecycle of an experiment on the NorNet / MONROE platform.

We will create a simple experiment (i.e. ping www.google.com 10 times), run it on the platform and finally retrieve the results.

The lifecycle of an experiment in brief

The steps we have to follow from experiment definition to having results are:

1. We first "pack" our scripts into a container.
2. If needed we prepare some server side scirpts. This is needed if, for example, our experiment follows the client-server paradigm.
3. We use some helper scripts to interface with the scheduler and define on which nodes and when the container will be deployed and run. Alternativly we can use the web interface of the scheduler at https://haugerud.nntb.no.
4. We retrieve the results from the temporary storage server pioneer.nntb.no.

Repository structure

• simpleping: The directory of the container. We must always use the name of the container as the name of is directory, because the scripts use the directory name to name to tag and push the container to dockerhub. Here the container is called simpleping, so we also name the directory simpleping. Do not use capital letters.
• scheduling: Documentation and scripts on how to interact with the scheduler. The README.md files explains how to use the API. There are also some helper scripts to interact with the scheduler programmatically.
• nodeDevelopment: instructions on how to connect directly on a node to test your containers.
• nodeResults: Directory where we fetch the files created by the experiements. It includes the files created by the container as well as as dignostic files created automatically, such as the deployment log and a log with everything printed on the screen. Each folder is an expriment instance run and is fetched "as is" from the temporary storage server pioneer.nntb.no.