Once the nodes are configured, it is time to set up the experiment. We will start by allocating a new slice named
bit-experiment. First, open the slices page by clicking on the top menu
Slices. You will find a list with two slices registered, but not deployed. They are created by default each time that the VCT container (VCT-C) is initialized.
For more information about the status of slices and slivers, check their wiki page.
Additionally to the slices' name and their set status, the list also shows the expiration date and the group to which the slices belong. Remember that only the members that are part of the slice's group have the permission to create, modify or delete them.
To allocate a new slice, click on the
Add slice button. The slice creation form will require some basic information about your new slice.
Nameis a unique name that the system will use to refer to your slice.
Descriptionis an optional field. However, we strongly encourage experimenters to add a small description of the slice purpose in this field. It will help testbed and node administrators identify problems early.
Templatefield allows researchers to select the base Linux distribution to use for their slivers. Current options are: OpenWRT-based and debian-based systems. Later we are going to see how to select a different system template for each of our slivers, but for now just select the debian system by selecting the
exp_dataallows to provide a compressed file with experimental data and initialization scripts. We are going to cover it later, but by now select the
Set stateis the field that sets the desired state for the whole slice. As at this point we haven't deployed the slice yet (meaning, that our slice doesn't have any slivers), thus we cannot change its state.
The advanced options will be covered in the third part of this tutorial.
Finally, click on the
Save and continue editing button to save your changes and start adding some slivers.
Now if you click on the
Add slivers button of the Slivers section of the slice configuration window, you can see a list of all the nodes that can currently run one more slivers. It includes all the nodes in PRODUCTION state, independent of the group it belongs to. In our case, we have access to all the virtual devices previously configured. There are, however, two limitations:
Select all of the nodes and click on the
Create slivers button.
This will bring up the connectivity configuration screen.
As we need public IPv4 connectivity between the slices, activate the
The isolated connectivity option cannot be chosen at this point. If such an option is desired, the slice needs to be assigned a VLAN ID and the slivers need to be configured one by one by adding a new interface of the type isolated and specifying a parent interface for it.
Yes, create slivers to confirm the operation. It will send you back to the previous site.
This step finishes the allocation process, which means that the controller has created all necessary configuration files for each sliver. However, nodes will not do anything until we try to deploy the slivers on them.
To do it, change the state to DEPLOY and confirm clicking on the
The new set state of the slice will change to DEPLOY. If you want to verify the current status of the slivers, use the top menu to access sliver details (Slice → Slivers).
The next screen will show the list of slices configured and some useful information.
State linkare the desired and current state of each sliver.
Public4are the number of network interfaces connected to the management network.
Managementindicates if the controller can contact the sliver, meaning that new configuration will be changed without problems.
Note that sliver name includes the slice where it belongs and the node in which it will be deployed. By clicking on the name it is possible to change the configuration for a particular sliver. Default configuration is the configuration of the slice.
Remember that the testbed operation architecture is based on nodes pulling configuration files. It means that you will need to wait for a while before some operations are completed. The waiting time can be in the order of minutes, and is generally shorter in the VCT container than on the real testbed.
When all slivers are deployed, set the slice status to START.
Then wait until all slivers are running.