OSPFN Manual
v0.2.0

Copyright (c) 2012, University of Memphis and University of Arizona.
Release Date: 05/11/2012

OSPFN is a routing software package for Named Data Networking (NDN) based 
on Quagga.  This manual covers OSPFN v0.2.0.

1. About OSPFN
2. Distribution 
3. OS Support
4. Requirements
5. Installation
6. Configuration
7. Running
8. Checking OSPFN
9. OSPFN Log
10. Stopping ospfn
11. Stopping ccnd, zebra, ospfd
12. Startup Scripts
13. Contributors



==================================
1. About OSPFN
==================================
OSPFN has been developed by the University of Memphis and the University of 
Arizona for supporting the paradigm shift in networking from IP to Named 
Data Networking (NDN).  It is an adaptation of the OSPF protocol implemented 
using the open source routing suite Quagga.  OSPFN uses OSPF's Opaque LSA 
capability to announce name prefixes and update the FIB of CCNx.  OSPFN 
is in its early stage of development. Feedback and bug reports are welcome.


==================================
2. Distribution
==================================
Currently OSPFN v0.1.0 is distributed by the University of Memphis and the
University of Arizona. Technical questions should be directed to 
Cheng Yi <yic@EMAIL.ARIZONA.EDU> and A K M Mahmudul Hoque <ahoque1@memphis.edu>. 
Source code can be downloaded from github by running the following:

$ git clone git://github.com/NDN-Routing/OSPFN2.0.git


==================================
3. OS Support
==================================
The current release was developed and tested on Ubuntu 10.04, 10.10, 11.04, FreeBSD 9.0 and Fedora 9. Other Unix variants may also work, but have not been tested yet.

==================================
4. Requirements
==================================
CCNx needs to be installed first.  CCNx can be obtained from
http://www.ccnx.org/software-download-information-request/download-releases/. 

Requirements for CCNx as well as installation instructions can be found in 
the README file in the CCNx package.  Typical installation steps for CCNx 
are as follows:

$ ./configure
$ make
$ sudo make install

CCNx v0.4.0 was used for testing and development.


==================================
5. Installation
==================================
Check out the OSPFN source code.  You should now have a directory "OSPFN"

Before running configure, the permissions on the OSPFN may need to be changed. Once that has been done, run configure with the following options:

$ ./configure --enable-opaque-lsa --disable-ipv6 --disable-ripd --disable-ripngd --disable-ospf6d --disable-bgpd --disable-bgp-announce --sysconfdir=/usr/local/etc/quagga --localstatedir=/var/run/quagga-state

Notes:

(a) here we specify the quagga configuration and state file directories in 
configure using the --sysconfdir and --localstatedir options, respectively. 
We assume /usr/local/etc/quagga and /var/run/quagga-state are used.
You can specify other directories.

(b) the --disable options are to avoid compiling Quagga components that we 
do not need.

A user named quagga needs to be created:

$ sudo addusr quagga

Next we need to create two directories: /usr/local/etc/quagga 
and /var/run/quagga-state (or any directories you specified in configure).
Both directories must be readable and writable by the user quagga and 
group quagga.

$ cd /usr/local/etc/
$ sudo mkdir quagga
$ sudo chown quagga:quagga quagga
$ sudo chmod g+w quagga
$ cd /var/run
$ sudo mkdir quagga-state
$ sudo chown quagga:quagga quagga-state
$ sudo chmod g+w quagga-state

Now we can compile and install the programs.

$ cd OSPFN
$ make
$ sudo make install


Note: In some OS /var/run is not persistent ie after restart /var/run/quagga-state directory disappear. To get around this problem one can create directory in persistent partition and change localstatedir=/var/run/quagga-state to
localstatedir = /directory/path/to/quagga-state. Or someone can also make /var/run/qugga-state persistent by setting appropriate permissions in /etc/rc.local.

==================================
6. Configuration
==================================

6.1 Get IP Addresses for GRE Tunnels

On the NDN testbed, each site is allocated a /24 block out of 10.0.0.0/8 
for setting up GRE tunnels for OSPFN traffic.  Please contact Colorado 
State University about which /24 block your site can use. Within each site, 
you can allocate addresses from your site /24 block to multiple tunnels 
or multiple OSPFN nodes as you wish.


6.2 Set Up GRE Tunnels

If two OSPFN routers are not physically connected, a GRE tunnel between 
the routers is needed.  Note that you need to make sure your firewall 
allows GRE traffic to pass through.

We use the following topology to show how to set up a GRE tunnel
between netlogic.cs.memphis.edu and hobo.cs.arizona.edu.  Netlogic's 
public IP is 141.225.11.150 and private tunnel IP is 10.0.1.10.
Hobo's public IP is 150.135.82.77 and private tunnel IP is 10.0.1.9. 


Public IP: 141.225.11.150                    Public IP:150.135.82.77
________________________  10.0.1.10  10.0.1.9 ___________________
|netlogic.cs.memphis.edu |___________________|hobo.cs.arizona.edu|
------------------------                      -------------------


6.2.1 GRE Tunnel Configuration in Ubuntu

Install the GRE module:

$ sudo modprobe ip_gre

Add the following to /etc/network/interfaces on netlogic.cs.memphis.edu:

auto netlogictohobo 
netlogictohobo inet static
address 10.0.1.10
netmask 255.255.255.252
pre-up iptunnel add netlogictohobo mode gre local 141.225.11.150 remote 150.135.82.77 ttl 255 
pointopoint 10.0.1.9
up ifconfig netlogictohobo multicast

Note that "netlogictohobo" is the name that you give to the tunnel interface.

Similary, add the following to the /etc/network/interfaces file on hobo.cs.arizona.edu:

auto hobotonetlogic
iface hobotonetlogic inet static
address 10.0.1.9
netmask 255.255.255.252
pre-up iptunnel add hobotonetlogic mode gre local 150.135.82.77 remote 141.225.11.150 ttl 255 
pointopoint 10.0.1.10
up ifconfig hobotonetlogic multicast

Restart networking:
$ sudo /etc/init.d/networking restart

The command "ifconfig -a" should list the new tunnel interface.  If the 
GRE tunnel is set up correctly, pinging the other end's private IP should work.


6.2.2 GRE tunnel configuration in Fedora

On netlogic.cs.memphis.edu:

Create a file /etc/sysconfig/networking/devices/ifcfg-netlogictohobo, 
with content:

DEVICE=netlogictohobo
TYPE=GRE
ONBOOT=yes
MY_INNER_IPADDR=10.0.1.10
MY_OUTER_IPADDR=141.225.11.150
PEER_INNER_IPADDR=10.0.1.9/30
PEER_OUTER_IPADDR=150.135.82.77
BOOTPROTO=none
TTL=64

Create hard links /etc/sysconfig/network-scripts/ifcfg-netlogictohobo and 
/etc/sysconfig/networking/profiles/default/ifcfg-netlogictohobo 
to /etc/sysconfig/networking/devices/ifcfg-netlogictohobo

Bring up the interface:
$ sudo ifup netlogictohobo

Similar configurations should be created on hobo.cs.arizona.edu.

The command "ifconfig -a" should list the new tunnel interface.  If the 
GRE tunnel is set up correctly, pinging the other end's private IP should work.


6.2.3 GRE tunnel configuration in FreeBSD

On netlogic.cs.memphis.edu:

Run the following command to create a new gre tunnel:

# ifconfig gre create

  gre1
  
It will create a new gre tunnel and display the name of it. In this example, the
 name is
gre1.

Now run the following commands to configure the tunnel:

# ifconfig gre1 10.0.1.10 10.0.1.9 netmask 255.255.255.252

# ifconfig gre1 tunnel 141.225.11.150 150.135.82.77

To verify the tunnel setup on netlogic, run the following:

# ifconfig gre1
gre1: flags=9051<UP,POINTOPOINT,RUNNING,LINK0,MULTICAST> metric 0 mtu 1476
        tunnel inet 141.225.11.150 --> 150.135.82.77
        inet6 fe80::211:43ff:fec2:61e0%gre1 prefixlen 64 scopeid 0xa
        inet 10.0.1.10 --> 10.0.1.9 netmask 0xfffffffc
        nd6 options=29<PERFORMNUD,IFDISABLED,AUTO_LINKLOCAL>
		
Now we need to make the tunnel permanent, we need to add the appropriate configu
ration
to the /etc/rc.conf configuration file. Tunnel configuration for netlogic in the
 /etc/rc.conf
file looks like the following code:

cloned_interfaces="gre1"
ifconfig_gre0="inet 10.0.1.10 10.0.1.9 netmask 255.255.255.252 tunnel 141.225.11
.150 150.135.82.77 up"

If there are multiple gre tunnels, we specify the name in cloned_interfaces sepa
rated by
a space e.x. cloned_interfaces="gre1 gre2" and add the ifconfig line for the new
 gre tunnel
below the previous one.

Similar configurations should be created on hobo.cs.arizona.edu. 

If the GRE tunnel is set up correctly, pinging the other end's private IP should
 work.


6.3 Quagga Configuration

The OSPFN source code includes the Quagga 0.99.17 code, so no separate 
installation of Quagga is needed, but you do need configure it. Instructions 
are available at http://www.quagga.net/docs.php.

By default, the Quagga configuration file ospfd.conf is located at 
/usr/local/etc/, but we have specified our own directory 
/usr/local/etc/quagga.  The user should copy ospfd.conf.sample to ospfd.conf 
in the same directory and modify the ospfd.conf file as necessary.
Same for zebra.conf.

A sample configuration file (ospfd.conf) for Quagga OSPF would look like 
below for netlogic.cs.memphis.edu:

!
! OSPF configuration saved from vty
!   2011/09/07 14:12:35
!
hostname netlogic.cs.memphis.edu 
password pwd
enable password pwd 
log file /var/log/quagga/ospfd.log
!
!
!
interface netlogictohobo
 description link to hobo.cs.arizona.edu
!
access-list ospfn permit 10.0.0.0/8
access-list ospfn deny any
!
router ospf
 ospf router-id 141.225.11.150 
 redistribute connected
 distribute-list ospfn out connected   
 network 10.0.1.8/30 area 0.0.0.0 
 capability opaque
!
line vty
!
!end of configuration file here

--------------------------------
Notes:

1. We are using the "redistribute connected" command to make the OSPF 
router an ASBR, so that we can retrieve the next hops to other routers 
from OSPFd. This is a temporary work-around until we find a better 
solution.  To prevent potential looping caused by the "redistribute 
connected" command, we are filtering out any prefixes that do not belong 
to the NDN testbed using the "access-list" commands, which only 
permit routes to 10.0.0.0/8 networks to be advertised. 

2. Router id: we are using the public IP of each router as its router 
ID to ensure uniqueness across the whole network.

3. The log directory (/var/log/quagga in the example) must be readable 
and writable by user quagga and group quagga.


6.4 OSPFN Configuration

By default, OSPFN looks for the configuration file ospfn.conf in the 
local directory.  Users can also supply a configuration file name when 
starting OSPFN. Configuration commands are entered one command per line 
into the configuration file.  Currently OSPFN supports two commands, 
ccnname and logdir. The ccnname command specifies name prefixes 
(in URI format) advertised by the router. The logdir command specifies a 
directory to write log files to.

1. ccnname name_prefix op_id

name_prefix : a name prefix, in URI format, to be advertised by this router.
op_id : the desired ID for this name LSA.  The op_id must be unique among 
the name prefixes advertised by the same router. 

2. logdir dir

dir : directory to write log files to

A sample configuration file may contain the following commands:

ccnname /ndn/memphis.edu/netlab/ 1
ccnname /ndn/memphis.edu/cs/ 2
logdir /var/log/ospfnlog/

Note that the log directory must be readable and writable by the user
that runs ospfn.

3. multipath-order a.b.c.d pre_order

Configure ospfn to add multipath for prefixes announced by other routers based
on the pref_order. Higher the value higher the preference.But highest preference
is given to the actual next hop from ospfd calculation. As an example if 
configuration has following line for multipath-order

multipath-order a.b.c.d 30
multipath-order x.y.z.1 15
multipath-order x.y.z.9 40

ospfn get nexthop for /ndn/example/prefix is a.b.c.d no ospfn will add fib entry
in the following order

/ndn/example/prefix x.y.z.1
/ndn/example/prefix x.y.z.9
/ndn/example/prefix  a.b.c.d

NB: ccnd give highest preference to the face added last.

==================================
7. Running OSPFN
==================================

First start zebra so that ospfd can acquire needed interface information. 
Next start ospfd and ccnd, and then start ospfn.  The commands needed are 
shown below but one can just run the script startAllFourDaemons.sh located
in script folder to run all four daemons.

$ sudo zebra -d
    
$ sudo ospfd -d -a
    
$ ccndstart   
    
$ ospfn



ospfn can also take several command-line options:

-f/F file_name: specify the configuration file.  
    (if this option is not included, OSPFN will look for the default 
     file named ospfn.conf in the current directory)
-d: OSPFN will run in daemon mode.
-h: OSPFN will print out the usage information for ospfn running options
-n: OSPFN will not write log information

Before running ospfn, you may need to give your user write access to the quagga-state directory. This can be done with the following command:

$ sudo usermod -a -G quagga username


==================================
8. Checking OSPFN
==================================
One can check whether OSPFN is working correctly by checking the FIB 
entries of the nodes in the network.  For example, assume 
netlogic.cs.memphis.edu and hobo.cs.arizona.edu are running OSPFN:

Netlogic declares the following name prefixes in its configuration file:

ccnname /ndn/memphis.edu/netlab/netlogic 1
ccnname /ndn/memphis.edu/netlab/ospfn 2

Hobo declares the following name prefixes in its configuration file:

ccnname /ndn/arizona.edu/hobo/name1 1
ccnname /ndn/arizona.edu/hobo/name2 2

If OSPFN is running successfully, Netlogic's advertised name prefixes 
located in its configuration file should be found in Hobo's FIB. 
Likewise, Hobo's advertised name prefixes should be found in Netlogic's 
FIB. Each FIB can be checked in a browser at http://<DNS name or address>:9695 
(eg: http://netlogic.cs.memphis.edu:9695) or http://localhost:9695.

A node's OSPFd routing table can also be checked to see if it has correct 
routes to all other router's address prefixes:

(1) connect to ospfd by typing the following command in terminal:

$ telnet localhost ospfd

(2) provide password configured in ospfd.conf

(3) type the following command:
    
$show ip ospf route


==================================
9. OSPFN Log
==================================

OSPFN creates a detailed log for every run of OSPFN in the directory 
configured by the ospfn configuration file.  Log file 
names are of the format yyyymmddhhmmss.log.  If the "logdir" 
command is not specified in the configuration file, OSPFN will
by default create a directory named ospfnLog in the user's home 
directory and write the log files there.


==================================
10. Stopping ospfn 
==================================

For stopping ospfn type ospfnstop in terminal.

$ospfnstop

ospfnstop will take a while to get response from ospfn and delete all the
fib entries by ospfn. After deleting the fib entries ospfn will stop and
ospfnstop will exit itself. 

A precautious measure is to always restart all 4 daemons: zebra, ospfd, ccnd, ospfn.
Running restartAllFourDaemons.sh script in script folder will restart all four
daemons.

==================================
11. Stopping ccnd, zebra, ospfd 
==================================

For stopping ccnd, zebra, ospfd and ospfn one can follow the below process or can 
invoke the stopAllFourDaemons.sh script located in script folder to stop all 
processes.

Stopping ccnd: stop ccnd by typping into terminal 
 $ccndstop

Stopping zebra: Kill zebra process by finding its process id, One can find process id 
by typing into terminal 
 $ ps -eaf | grep zebra

Then you will find result like below 

quagga    2786     1  0 Oct09 ?        00:00:00 zebra -d

After that type into terminal
 $kill 2786

Stopping ospfd: Kill zebra process by finding its process id, One can find process id 
by typing into terminal 
 $ ps -eaf | grep ospfd

Then you will find result like below 

quagga    2788     1  0 Oct09 ?        00:00:00 ospfd -d -a

After that type into terminal
 $kill 2788

Stopping ccnd: Run ccndstop from terminal to stop ccnd
 $ccndstop

==================================
12. Startup Scripts
==================================

For restarting all four process at once script restartAllFourDaemons.sh is provided
in script folder. Invoking of this script will restart all four process ( zebra, 
ospfd, ccnd and ospfn). This restarting script uses two other script for stopping:
stopAllFourDaemons.sh and starting:startAllFourDaemons.sh. User also can separately
use these two script to stop and start all four processes at once.

==================================
13. Contributors
==================================

Implementation and Testing: Cheng Yi (U. Arizona), 
A K M Mahmudul Hoque (U. Memphis), Yaoqing Liu (U. Memphis), 
Gus Sanders III (U. Memphis)

Advising: Lan Wang (U. Memphis), Beichuan Zhang (U. Arizona).