OneLab - Future Internet Testbeds

Installing nepi-ng

What is nepi-ng ?

The commands that we have seen in the previous tutorial are useful for making quick changes on the testbed.

However once you are set up, it is much more convenient to control experiments right from your own laptop, and to this end we provide a set of tools that we collectively refer to as nepi-ng.

nepi-ng = asynciojobs + apssh

nepi-ng is simply made of 2 python libraries. This page gives you detailed instructions for installing them. You will also find a more complete documentation on these 2 tools below :

Terminal-based

All the tools that we will write will run in a terminal, we assume you are familiar with dealing with such interfaces.


Requires asyncio and python-3.9

nepi-ng is a set of python libraries; these are based on the asyncio library; these libraries require *at least python-3.9.

$ python --version
Python 3.9.13

Installing asynciojobs and apssh

You need to install these libraries, the recommended way is through pip3 like this - if you're not using virtualenv or anaconda, you might need to run these commands under sudo:

[sudo] pip3 install apssh

This should be sufficient as asynciojobs is required by apssh. You can check the installed versions like this

$ python3 -c 'from asynciojobs import __version__; print(__version__)'
0.13.1
$ python3 -c 'from apssh import __version__; print(__version__)'
0.14.1

Note that if you need to upgrade in the future, you will need to run

[sudo] pip3 install --upgrade apssh

and that will bring you the latest release published on pypi.org.

Optional - installing the r2lab python library

It is recommended to also install the rl2ab python library, that we will briefly touch on in this tutorial in the D series. The library is also required for running some of the demos:

[sudo] pip3 install r2lab

and then likewise

$ python3 -c 'from r2lab import __version__; print(__version__)'
0.1.1

ssh and ssh-agent

nepi-ng does not require a native ssh client to be installed on your laptop, as it uses the great asyncssh library instead. However, it is recommended to have one installed, if only for running an ssh agent, as using ssh repeatedly is almost impossible without an ssh-agent.

You will see that in all the code of these tutorials, there is no option to specify an alternate key. This is because all the examples in this tutorial use the set of keys known to your ssh agent.

You can inspect and manage this set of keys using the ssh-add command - at least on unix-based systems.

# inspect current contents
~/.ssh $ ssh-add -l
1024 SHA256:KNm0U4SgFV9bY957hJAIRR68n5AZHQ6e1gbiXrGHuOA /Users/parmentelat/.ssh/id_rsa (RSA)

# if the key is not present in the list, just add it
~/.ssh $ ssh-add path/to/the/key/id_rsa.pub

# check for the addition
~/.ssh $ ssh-add -l
1024 SHA256:KNm0U4SgFV9bY957hJAIRR68n5AZHQ6e1gbiXrGHuOA /Users/parmentelat/.ssh/id_rsa (RSA)

ssh-agent lifespan

In your ssh agent, the keys you add are kept in memory. This means that they can only be added for the lifespan of the agent, and need to be added again if, for example, you log out of your session, or otherwise reboot your laptop.

other means to deal with keys

The reasons we use keys from an ssh agent in the tutorial are:

  • first it makes code a little simpler

  • second, first in standard situations a private key is password protected. And when a private key indeed is password-protected, then using an ssh agent really becomes almost mandatory, because otherwise you get prompted your password every single time you run an ssh command.

However if you prefer, you can as well run without an ssh agent, and tweak your script so that a specific key be used instead.

xxx we cannot give details at this point because it is too early; to be added somewhere else towards the end.