Version 15 (modified by, 10 years ago) (diff)


This page is still under construction, please email any comments/corrections you have to

Get 2 ProtoGENI hosts connected through OpenFlow at the GPO lab

Using GPO credentials

In order to grab all the resources needed, you will need to use the Omni client, which is a software that uses the GENI AM API to get resources from the different Aggregate Managers.

You will also need to have GENI credentials from GPO's ProtoGENI Clearinghouse. Please visit the 'Sign Me Up' wiki page to get more information.

Once you have your GENI credentials and you have configured Omni you are ready to go through this example.

1. Create a slice and renew it

nriga@pella:~/gcf$ ./src/ createslice niky-pgof-gpo

nriga@pella:~/gcf$ ./src/ renewslice niky-pgof-gpo  20110520T00:00:00Z

The default expiration time for a slice is only a few hours, so make sure that you renew your slice before adding any resources, since any resources that you acquire can not have an expiration time that outlives your slice.

2. Create a PG sliver and renew it

Our ProtoGENI aggregate has 11 nodes that are identically configured. In this example the provided rspec requests for any two ProtoGENI hosts. You can modify the rspec to request for more nodes if needed. You do not need to specify the topology in the ProtoGENI rspec file, since you are going to use OpenFlow to control your traffic and create your topology. Find out the available resources using the command listresources

nriga@pella:~/gcf$ src/ listresources -n -a

If there are enough available resources then create and renew your sliver.

nriga@pella:~/gcf$ src/ createsliver -n -a  niky-pgof-gpo pg.rspec

nriga@pella:~/gcf$ src/ renewsliver -a -n niky-pgof-gpo 20110520T00:00:00Z

For ProtoGENI slivers the expiration time is the same as the expiration time of the sliver, but you might want to make it a habit to renew your sliver after creating it since this is not true for all aggregates.

Use the sliverstatus command to check whether your resources are ready for use. Be patient it will take several minutes until your nodes boot and are ready for use. The sliverstatus will also tell you which hosts you have reserved. The hosts in the GPO lab are named pc1-pc11. Make a note of the number for each one of the hosts, since you are going to need it to configure your topology.

nriga@pella:~/gcf$ src/ sliverstatus -a  niky-pgof-gpo

When the sliverstatus says ready login to the nodes :

nriga@pella:~/gcf$ ssh -i /home/nriga/.ssh/gcf_id_rsa

You will need to follow these instructions in all your hosts in order to configure the network interface that is directly connected to the OpenFlow switch.

3. Create an OpenFlow sliver

Assuming that you will start with a local test only, you will need to control all the traffic coming and going to your hosts. When you need to connect this testbed to other sites, please let us know and we will give you further instructions on how to accomplish this. Take a look in the instructions about how to write OpenFlow rspecs.

You will need to modify the attached rspec with your own information. In particular you will need to:

  1. modify sliver description and email to match your own information
  2. point your sliver to your own controller. In the example rspec, the OpenFlow resources point to a controller running at a host in the GPO lab ( listening on port 1716 for connections. You can use any host you want for the controller as long as it is reachable through the internet. Performance-wise you might want to use one of your ProtoGENI hosts (make sure you specify the control interface of the host, e.g. as your OpenFlow controller. If you have never used OpenFlow before, let us know and we can help you get started. iii Change the port numbers that you request based on the hosts you got. The port number you need to use is port=24+xx, where xx is the number from the hosts you have reserved, the example rspec is for the case you reserved pc4 and pc6 (ports 28 and 30).

To get the OpenFlow resources :

nriga@pella:~/gcf$ src/ createsliver -n -a pg-of niky-pgof-gpo of.rspec 

To test connectivity you can first start your controller acting as a Layer 2 switch and try pinging between your hosts.

Configure PG Hosts

Login to the nodes :

nriga@pella:~/gcf$ ssh -i /home/nriga/.ssh/gcf_id_rsa

On each of the PG hosts you reserved do:

  1. Add these lines in the end of /etc/network/interfaces, edit this file as sudo:
    auto eth3.1750
    iface eth3.1750 inet static
        mtu 1500

where xx = the number of the pc host, xx=04 for pc4 and xx=06 for pc6.

  1. Install the vlan package
     sudo apt-get install vlan
  2. Reboot the machine
    sudo init 6 && exit

When the nodes reboot they will have a new interface eth3.1750 that is configured in the subnet.

Attachments (3)

Download all attachments as: .zip