OneLab - Future Internet Testbeds

Shell tools in the gateway

Logging in the gateway

Once you have obtained a slice account on R2LAB (faraday), you can reach the R2lab gateway using ssh

$ ssh

If by any chance your public key is not at its standard location, then place -i option in the command line and inform the path of it.

Listing commands

From your bash account on the gateway, you have a few very simple but handy tools at your disposal for the early steps of your experiment, like seeing the nodes status, turning them on or off, and loading images.



From the gateway, you can use the following hostnames to refer to nodes

  • fit08 : refers to the control wired interface on node 8; the control interface is configured in all our images to start up automatically at boot-time so the nodes can be reached.

  • data08 : refers to the data wired interface; this one is not automatically turned on, it is up to you to use it or not, you can use DHCP for that purpose.

  • reboot08 : refers to the ethernet interface of the CMC device on the node, that allows for remote management of the motherboard (i.e. turning nodes on and off)

Here's an example of how these names resolve. Beware that the IP address of the reboot interface might occassionnally not be directly to the node index, but this is seldom used by experiments.

your_slicename@faraday:~$ host fit08
fit08 has address
your_slicename@faraday:~$ host data08
data08 has address
your_slicename@faraday:~$ host reboot08
reboot08 has address

Checking leases

In case you're unsure about the current status of reservation, you can list reserved timeslots - known as leases - with


or check that you currently hold a reservation with

rleases --check

This is a poorman's tool, as of course the scheduler will give you that same information in a much nicer way.

Note also that this is in fact equivalent to

rhubarbe leases

rhubarbe (see it on github) being the set of tools that help us run the testbed. Indeed many of the convenience functions starting with r in fact are aliases to a rhubarbe subcommand.

Switch off the testbed when you're done

One very frequently used command, that requires that you have a valid lease - or at least that no lease is currently active - is the one that turns off the nodes when you are done:


which is an alias for either

rhubarbe bye

or simply


Please try to make sure to use it, especially when you have many nodes running.

In the next tab we will see how we can focus on a specific set of nodes, and easily control them.<

Monitor what you're doing

Be aware that you can have a global and live view of the testbed status right in the r2lab website. This link works for all visitors, and if you are logged in you can also see this page that will also show you the state of the reservation system.

Selecting nodes

Most of the time, you will want to manage a selected subset of nodes. There's a simple mechanism in place so you don't need to specify your nodes for each and every command, by defining the environment variable NODES. For this the nodes command is your friend

To select nodes, use the nodes command. To select nodes 1 2 4 5 33 and 37 you could do this (~ stands for negation)

nodes 1-5 33,37 ~3

To select all nodes, you could do


To remove all nodes between 3 and 35 from your selection; same with nodes-add for adding nodes

nodes-sub 3-35

To see your selection, just run


Finally the commands nodes-save and nodes restore let you name selections, and then reinstate them

nodes-save run1
nodes-restore run1

Are these nodes on or off


By default - i.e. with no argument - this command and most of the ones we will show here operate on your nodes selection, but you can always specify another set of nodes to operate on, regardless of the overall selection

So this will give you the status of nodes 1 2 and 3, no matter what you have selected

st 1-3

Managing nodes (turning them on or off, or rebooting)

To turn on your selected nodes selection just do


Or again, if you want to turn on node 3 only, just do

on 3

Turning them off is of course just


You can trigger a reset (reboot) on a node - provided it is already on, with


To see the list of nodes that are ON


You can select all the nodes currently ON with

focus-nodes-on -a

To see the linux version running in the nodes (this is less sophisticated than what the livetable would provide)


Loading images

The tool for loading images is called rload. It is in fact a shortcut for rhubarbe load, like most commands described here

rhubarbe --help

See the source code for rhubarbe for more details.

Back to image loading, you will first want to know which images are available:


Assuming you want to load the latest fedora image, you would just do

rload -i fedora

that would act on all your selected nodes (or as always add a list of nodes to the command)

rload -i fedora 1-10

image loading has a fancier mode that can come in handy for troubleshooting: the --curses mode - -c in short - gives you a live progressbar for each node. Note however that the curses mode is not suitable for scripting, as it will wait for a keystroke before exiting.

Waiting for nodes

You can wait for all the selected nodes to be ssh-ready by running


This command, like all the rhubarbe-related commands, has a default timeout, that you can change with (-t is the shortcut)

rwait --timeout 30

ssh-ing into nodes

Once you have loaded an image, you can enter all nodes by just doing

ssh root@fit25

or just, if you're really lazy

ss 25

You can run a command on all selected nodes with

map ip addr show

this time of course, you cannot specify another set of nodes than the selection.

Saving images

You have the ablility to save the image - this now of course applies only to one node at a time. To save node 25

rsave 25 -o my-imge-name

This ends up in the common repository /var/lib/rhubarbe-images, under a name that holds the hostname and saving time. You can also provide an extra name to rsave with the -o option.

Images that may be of common interest usually needto be renamed; get in touch with the admins if you need to do this.


For now we have two commercial phones available right in the room; each phone can be controlled through a dedicated MAC box, called macphone1 and macphone2.

As far as shell commands are concerned, since that is the focus of this tutorial, be aware that you can reach e.g. the second macphone from faraday by doing just


and from then, as usual


to get a reminder.

Please refer to this page for more details on this offering, and how to manage these phones e.g. through a VNC session.