c-bgp-users — Discussion for users of c-bgp

Re: [C-bgp-users] C-BGP decision process modification: inconsistent AS path

[Bruno Q. <bru...@um...>]

Hi Yaping,

> hi Bruno,
> Thanks a lot for the quick response. I followed your suggestions to 
> trace through the decision process using the "debug dp" command, and 
> found the bug in my code :-)

Great. That "debug dp" command also saved my life a few times ;-)

> May I ask another question (should be the last one...)? I'm planning 
> to evaluate the incremental deployment of our proposed protocol (say 
> BGP*): by having one AS running BGP* decision process, and the rest 
> running BGP by default. In order to do this,
> - in bgp/dp_rules.h/c: I defined another set of rules in DP_RULES_NEW[]
> - in bgp/as.c: I selectively choose the decision process
>   if(router->asn matches the AS selected to deploy)
>     call DP_RULES_NEW
>   else
>     unchanged.

I would suggest that you add an option in each router that determines 
which set of decision process rules should be applied.

This can be done with a small change in src/bgp/types.h
* look for structure bgp_router_t { ... }
* insert your option, e.g. "int dp_type;"   (or use an "enum" type)
* add a CLI command that allows to change that variable + base the 
selection of the dp rules on the value of this variable

> my question is: I want to define a new command (or I could also reuse 
> your commands, if that's easier), like "incremental deployment ASXXX". 
> I want to parse the AS number specified in the .cli file, store it in 
> some variable, and pass it to as.h/c.
> I'm digging into your code, but haven't figured out:
> - which file parses the .cli configuration. i'm looking into the /cli 
> directory, but haven't figured out.
> - how to pass the AS number to bgp/as.c, where do you store the 
> variables? any examples which you pass variables from config to bgp code?
> Could you help to point me to some related files (pieces of 
> code/examples)? Really appreciate your help here!

The CLI parsing code lies within the libgds library. However, the actual 
supported commands are defined in c-bgp under src/cli. The set of 
commands spans several files.

Let's say you want to add a command "bgp router ASxxx dp_type xxx". This 
can be done with a change in file src/cli/bgp.c
* to add a command, you need code that register the command and provide 
the implementation of the command.
* to register your command, look for function "_register_bgp_router": 
this function is responsible for registering the commands under "bgp 
router". Look at how other commands are added. You can find a doxygen 
documentation for the various libgds CLI commands within libgds's source 
archive.
* to provide the implementation of the new command, you need to provide 
a C function that takes well-defined parameters. Look at 
"cli_bgp_router_recordroute" for an example. This command has the 
following structure.

     int cli_bgp_router_recordroute(cli_ctx_t * ctx, cli_cmd_t * cmd)
     {
       bgp_router_t * router= _router_from_context(ctx);
       const char *   arg= cli_get_arg_value(cmd, 0);
         (...)
         implementation of command
     }

Cheers,
Bruno


> thanks,
> Yaping
>
> On Mar 22, 2011, at 4:39 PM, Bruno Quoitin wrote:
>
>> Hi Yaping,
>>
>> Are you sure there are no pending events in the simulator's queue 
>> (check with "sim queue show") ?
>> That might explain such inconsistencies.
>>
>> Did you check the current best path in every router along the path 
>> selected by A ?
>> Did you check the content of A's adj-rib-in (check with "bgp router A 
>> show adj-rib in * *")
>> You can also display what the decision process did on a router (say 
>> A) with "bgp router A debug dp <prefix>"
>>
>> I have no more idea based on your short description. As a last 
>> resort, you can send me your code (if that's not private) and I will 
>> try to reproduce the behaviour you observed.
>>
>> Cheers,
>> Bruno
>>
>> hi C-BGP developers,
>> My name is Yaping Zhu, and I'm a graduate student at Princeton 
>> University. I'm using C-BGP to evaluate a new routing protocol (which 
>> modifies the BGP decision process). However, I get inconsistent AS 
>> path seen from different vantage points. Would be really appreciated 
>> it you could help here.
>>
>> More specific, I change the BGP decision process in bgp/dp_rules.h/c 
>> in C-BGP to:
>> - disable the step of selecting route based on shorter AS path
>> - add a additional step, Prefer Recent Route: if the current route is 
>> of the same local-pref as the best route, then the decision process 
>> prefers current route.
>> After changing the code, I did the simulation to output the AS path 
>> from all the ASes. However, the AS path are not consistent. e.g. AS A 
>> is having B on it's path, say A X B Y Z; however, B is not selecting 
>> Y Z but W K as its AS path.
>> I checked my configuration, and the code I modified, which seems to 
>> be correct. Do you have any thoughts on what might go wrong?
>>
>> thanks,
>> Yaping
>


-- 
Prof. Bruno Quoitin                         Networking Lab
UMons, Belgium    http://informatique.umons.ac.be/networks
Phone: ++32 65 37 34 48             GSM: ++32 498 28 12 21

[C-bgp-users] old wiki and new attribute

[Dow S. <dow...@pr...>]

Hi Bruno.

I seem to remember there being an older c-bgp wiki with some  
additional documentation, but it looks like the redirect may have  
broken in the move to sourceforge?  Also, do you know of any examples  
and/or papers describing the implementation of an experimental (new)  
attribute in c-bgp?  (digging through the 2.0.0-rc3 code now).

Thanks,
Dow

Re: [C-bgp-users] C-BGP decision process modification: inconsistent AS path

[Bruno Q. <bru...@um...>]

Hi Yaping,

Are you sure there are no pending events in the simulator's queue (check 
with "sim queue show") ?
That might explain such inconsistencies.

Did you check the current best path in every router along the path 
selected by A ?
Did you check the content of A's adj-rib-in (check with "bgp router A 
show adj-rib in * *")
You can also display what the decision process did on a router (say A) 
with "bgp router A debug dp <prefix>"

I have no more idea based on your short description. As a last resort, 
you can send me your code (if that's not private) and I will try to 
reproduce the behaviour you observed.

Cheers,
Bruno

    hi C-BGP developers,
    My name is Yaping Zhu, and I'm a graduate student at Princeton
    University. I'm using C-BGP to evaluate a new routing protocol
    (which modifies the BGP decision process). However, I get
    inconsistent AS path seen from different vantage points. Would be
    really appreciated it you could help here.

    More specific, I change the BGP decision process in bgp/dp_rules.h/c
    in C-BGP to:
    - disable the step of selecting route based on shorter AS path
    - add a additional step, Prefer Recent Route: if the current route
    is of the same local-pref as the best route, then the decision
    process prefers current route.
    After changing the code, I did the simulation to output the AS path
    from all the ASes. However, the AS path are not consistent. e.g. AS
    A is having B on it's path, say A X B Y Z; however, B is not
    selecting Y Z but W K as its AS path.
    I checked my configuration, and the code I modified, which seems to
    be correct. Do you have any thoughts on what might go wrong?

    thanks,
    Yaping

[C-bgp-users] C-BGP decision process modification: inconsistent AS path

[Yaping Z. <yapingz@CS.Princeton.EDU>]

hi C-BGP developers,
My name is Yaping Zhu, and I'm a graduate student at Princeton  
University. I'm using C-BGP to evaluate a new routing protocol (which  
modifies the BGP decision process). However, I get inconsistent AS  
path seen from different vantage points. Would be really appreciated  
it you could help here.

More specific, I change the BGP decision process in bgp/dp_rules.h/c  
in C-BGP to:
- disable the step of selecting route based on shorter AS path
- add a additional step, Prefer Recent Route: if the current route is  
of the same local-pref as the best route, then the decision process  
prefers current route.
After changing the code, I did the simulation to output the AS path  
from all the ASes. However, the AS path are not consistent. e.g. AS A  
is having B on it's path, say A X B Y Z; however, B is not selecting Y  
Z but W K as its AS path.
I checked my configuration, and the code I modified, which seems to be  
correct. Do you have any thoughts on what might go wrong?

thanks,
Yaping

Re: [C-bgp-users] [cbgp] Filters in cbgp-2.0.0 rc

[Bruno Q. <bru...@um...>]

Hello,

Sorry for the late answer. Please also note that the official c-bgp 
mailing-list has moved to sourceforge (I cc:'ed the new mailing-list).

There is an updated documentation on-line available from the sourceforge 
web site (http://c-bgp.sourceforge.net/documentation.php). For 
documentation regarding the filters, please have a look at the following 
node: 
http://c-bgp.sourceforge.net/doc/html/nodes/bgp_router_peer_filter_add-rule.html
 From there, the "action" and "match" parts are described and the 
possible predicates and actions are also defined.

In cas you don't find the required information, please contact me again 
at my umons e-mail address.

Cheers,
Bruno


On 09/12/10 01:07, Palani kodeswaran wrote:
> Hello,
>
> Is the filter add-rule syntax the same in CBGP 1.4 and CBGP 2. Also, 
> where can I find documentation on filters for CBGP 2.
>
> Thanks,
> Palani.
>
>


-- 
Prof. Bruno Quoitin                         Networking Lab
UMons, Belgium    http://informatique.umons.ac.be/networks
Phone: ++32 65 37 34 48             GSM: ++32 498 28 12 21

Re: [C-bgp-users] Using cbgp simulator

[Yosef K. <yo...@ov...>]

Dear Prof Quoitin.
Thank you for your help. 
My aim is to use CBGP to simulate BGP network events and to export it to my
test environment. 
I would like to set the virtual peer as a rr-client and to record the routes
sent to it. 
>From documentation, virtual peers will not receive BGP messages. 
So I'll save router A rib and export it to my test environment.
Thanks Yosef Kahana

-----Original Message-----
From: Bruno Quoitin [mailto:bru...@um...] 
Sent: Wednesday, March 02, 2011 5:01 PM
To: Yosef Kahana
Cc: c-b...@li...
Subject: Re: Using cbgp simulator

Dear Yosef,

I copy the c-bgp mailing-list with my answers. They might be useful to 
other users.

On 01/03/11 13:37, Yosef Kahana wrote:
> Prof Quoitin.
> I'm using cbgp simulator and facing some difficulties.
> 1) simulated topology is:
> 5 BGP routers (a,b, and c have networks attached) 10.5.16.201-205 (a-e)
> the Links are a<->d
>                       b<->d
>                       b<->e
>                       c<->e
>                       d<->e
> I can see a and b networks at d
>                 b and c networks at e
>      but there is no propagation of networks between d and e.
>      there is no propagation of networks to a, b and c.
It looks like all the sessions are iBGP. A route received from an iBGP 
peer will not be redistributed to another iBGP peer.

You need to either
* establish a full-mesh of iBGP sessions among your routers. Note that 
you can achieve this quickly using the following command "bgp domain 
65024 full-mesh". Asa result, all routers in that domain will have iBGP 
sessions between each others;
* establish eBGP sessions;
* use route-reflectors: will allow propagation of iBGP learned routes 
between iBGP neighbors.


> 2) I would like to set a real BGP router as rr-client to d and e.
> their peer status is active but nothing is sent.
What do you mean by a real BGP router ? c-bgp will not establish BGP 
sessions with a real router.

However, you are allowed to define a "virtual" router in the simulation 
which will act as a placeholder for a real router. The purpose of 
virtual routers is to allow the injection of routes collected on a real 
router. In brief, this works as follows:

1). You collect routes on your real router X
2). You setup a BGP session between router A in c-bgp and virtual router 
X. To be clear, router A has been defined with "net add node" and "bgp 
add router" while router X has not been defined in c-bgp.
3). You set the peer X of the BGP session of step (2) as virtual

In the end, you should have something like

     bgp router A
         add peer AS(X) IP(X)
         peer IP(X) virtual
         ...
     bgp router A peer IP(X) recv "...MRT formated BGP route..."

> attached my configuration file.
By the way, I had to edit your file before testing because there were 
plenty of non-ASCII characters in the script. See for example the first 
line which has "" before the comment.

Regards,
Bruno


-- 
Prof. Bruno Quoitin                         Networking Lab
UMons, Belgium    http://informatique.umons.ac.be/networks
Phone: ++32 65 37 34 48             GSM: ++32 498 28 12 21

Re: [C-bgp-users] Using cbgp simulator

[Bruno Q. <bru...@um...>]

Dear Yosef,

I copy the c-bgp mailing-list with my answers. They might be useful to 
other users.

On 01/03/11 13:37, Yosef Kahana wrote:
> Prof Quoitin.
> I'm using cbgp simulator and facing some difficulties.
> 1) simulated topology is:
> 5 BGP routers (a,b, and c have networks attached) 10.5.16.201-205 (a-e)
> the Links are a<->d
>                       b<->d
>                       b<->e
>                       c<->e
>                       d<->e
> I can see a and b networks at d
>                 b and c networks at e
>      but there is no propagation of networks between d and e.
>      there is no propagation of networks to a, b and c.
It looks like all the sessions are iBGP. A route received from an iBGP 
peer will not be redistributed to another iBGP peer.

You need to either
* establish a full-mesh of iBGP sessions among your routers. Note that 
you can achieve this quickly using the following command "bgp domain 
65024 full-mesh". Asa result, all routers in that domain will have iBGP 
sessions between each others;
* establish eBGP sessions;
* use route-reflectors: will allow propagation of iBGP learned routes 
between iBGP neighbors.


> 2) I would like to set a real BGP router as rr-client to d and e.
> their peer status is active but nothing is sent.
What do you mean by a real BGP router ? c-bgp will not establish BGP 
sessions with a real router.

However, you are allowed to define a "virtual" router in the simulation 
which will act as a placeholder for a real router. The purpose of 
virtual routers is to allow the injection of routes collected on a real 
router. In brief, this works as follows:

1). You collect routes on your real router X
2). You setup a BGP session between router A in c-bgp and virtual router 
X. To be clear, router A has been defined with "net add node" and "bgp 
add router" while router X has not been defined in c-bgp.
3). You set the peer X of the BGP session of step (2) as virtual

In the end, you should have something like

     bgp router A
         add peer AS(X) IP(X)
         peer IP(X) virtual
         ...
     bgp router A peer IP(X) recv "...MRT formated BGP route..."

> attached my configuration file.
By the way, I had to edit your file before testing because there were 
plenty of non-ASCII characters in the script. See for example the first 
line which has "" before the comment.

Regards,
Bruno


-- 
Prof. Bruno Quoitin                         Networking Lab
UMons, Belgium    http://informatique.umons.ac.be/networks
Phone: ++32 65 37 34 48             GSM: ++32 498 28 12 21

[C-bgp-users] Updated documentation

[Bruno Q. <bru...@um...>]

Dear c-bgp users,

The c-bgp documentation has been updated.

-> An online command reference manual is available at 
http://c-bgp.sourceforge.net/doc/html.

-> A tutorial is available at http://c-bgp.sourceforge.net/tutorial.php

Cheers,
Bruno

Re: [C-bgp-users] Errors in cbgp-2.0.0-rc3

[Bruno Q. <bru...@um...>]

Thanks. Could you please try with libgds-2.0.0-rc3 and let me know how 
it goes.

Regards,
Bruno



On 09/01/11 16:46, Yosef Kahana wrote:
>
> Hi Bruno.
> I have a sabauon environment.
> GNOME Version 2.32.1, Build data 12/07/2010, Distributor Gentoo.
>
> libgds version: libgds-2.0.0-rc1. was downloaded with the distribution 
> of cbgp-2.0.0-rc3
>
> The first compiler error is:
> make[3]: Entering directory 
> `/home/yosefk/Tools/BGP/Simulator/BGP/cbgp-2.0.0-rc3/src/cli'
> /bin/sh ../../libtool  --tag=CC   --mode=compile gcc -DHAVE_CONFIG_H 
> -I. -I../../src -I..   
> -I/home/yosefk/Tools/BGP/Simulator/BGP/local/include        -g -O2 
> -Wall -O2 -MT libcli_la-sim.lo -MD -MP -MF .deps/libcli_la-sim.Tpo -c 
> -o libcli_la-sim.lo `test -f 'sim.c' || echo './'`sim.c
> libtool: compile:  gcc -DHAVE_CONFIG_H -I. -I../../src -I.. 
> -I/home/yosefk/Tools/BGP/Simulator/BGP/local/include -g -O2 -Wall -O2 
> -MT libcli_la-sim.lo -MD -MP -MF .deps/libcli_la-sim.Tpo -c sim.c  
> -fPIC -DPIC -o .libs/libcli_la-sim.o
> sim.c: In function 'cli_sim_options_loglevel':
> sim.c:130: error: too many arguments to function 'stream_str2level'
>
> Thanks Yosef.
>
> On 01/09/2011 04:35 PM, Bruno Quoitin wrote:
>> Hi Yosef,
>>
>> Could you tell me which version of libgds you are using. What is your 
>> build environment (OS + version).
>>
>> Then, what's the exact error message you received for the first error 
>> (message from compiler).
>>
>> Thanks.
>> Bruno
>>
>>
>> On 09/01/11 08:56, Yosef Kahana wrote:
>>> Dear Prof Quotin.
>>>
>>> When compiling cbgp-2.0.0-rc3 I got an error and sim.c line 130.
>>> The stream_str2level should receive char * and return the 
>>> stream_level_t.
>>> I fixed it as follow:
>>> line 130: if ((level = stream_str2level(arg)) < 0) {
>>>
>>>
>>> _After successful compilation the make check had returned:
>>> _yosefk@ov-dt-00116 ~/Tools/BGP/Simulator/BGP/cbgp-2.0.0-rc3 $ make 
>>> check
>>> Making check in src
>>> ......
>>> ......
>>> ......
>>> Testing: Net Network:network add subnet                         
>>> SUCCESS (0.0s)
>>> Testing: Net Network:network add subnet (duplicate)             
>>> SUCCESS (0.0s)
>>> Testing: Net Network:node send/bin/sh: line 5: 32114 Segmentation 
>>> fault      ${dir}$tst
>>> FAIL: cbgp-test
>>> ===========================================
>>> 1 of 1 test failed
>>> Please report to bru...@uc...
>>> ===========================================
>>> make[3]: *** [check-TESTS] Error 1
>>> make[3]: Leaving directory 
>>> `/home/yosefk/Tools/BGP/Simulator/BGP/cbgp-2.0.0-rc3/src'
>>> make[2]: *** [check-am] Error 2
>>> make[2]: Leaving directory 
>>> `/home/yosefk/Tools/BGP/Simulator/BGP/cbgp-2.0.0-rc3/src'
>>> make[1]: *** [check-recursive] Error 1
>>> make[1]: Leaving directory 
>>> `/home/yosefk/Tools/BGP/Simulator/BGP/cbgp-2.0.0-rc3/src'
>>> make: *** [check-recursive] Error 1
>>>
>>> Yosef Kahana, Senior Software Engineer, Oversi
>>> M: +972-52-3477562 | T: +972-77-333-7771 | F: +972-3-542-3165
>>>
>>> yo...@ov...  |www.oversi.com
>>
>>
>> -- 
>> Prof. Bruno Quoitin                         Networking Lab
>> UMons, Belgiumhttp://informatique.umons.ac.be/networks
>> Phone: ++32 65 37 34 48             GSM: ++32 498 28 12 21
>
>
> -- 
> Yosef Kahana, Senior Software Engineer, Oversi
>
> M: +972-52-3477562 | T: +972-77-333-7771 | F: +972-3-542-3165
>
> yo...@ov...  |www.oversi.com


-- 
Prof. Bruno Quoitin                         Networking Lab
UMons, Belgium    http://informatique.umons.ac.be/networks
Phone: ++32 65 37 34 48             GSM: ++32 498 28 12 21

Re: [C-bgp-users] Errors in cbgp-2.0.0-rc3

[Bruno Q. <bru...@um...>]

Hi Yosef,

Could you tell me which version of libgds you are using. What is your 
build environment (OS + version).

Then, what's the exact error message you received for the first error 
(message from compiler).

Thanks.
Bruno


On 09/01/11 08:56, Yosef Kahana wrote:
> Dear Prof Quotin.
>
> When compiling cbgp-2.0.0-rc3 I got an error and sim.c line 130.
> The stream_str2level should receive char * and return the stream_level_t.
> I fixed it as follow:
> line 130: if ((level = stream_str2level(arg)) < 0) {
>
>
> _After successful compilation the make check had returned:
> _yosefk@ov-dt-00116 ~/Tools/BGP/Simulator/BGP/cbgp-2.0.0-rc3 $ make check
> Making check in src
> ......
> ......
> ......
> Testing: Net Network:network add subnet                         
> SUCCESS (0.0s)
> Testing: Net Network:network add subnet (duplicate)             
> SUCCESS (0.0s)
> Testing: Net Network:node send/bin/sh: line 5: 32114 Segmentation 
> fault      ${dir}$tst
> FAIL: cbgp-test
> ===========================================
> 1 of 1 test failed
> Please report to bru...@uc...
> ===========================================
> make[3]: *** [check-TESTS] Error 1
> make[3]: Leaving directory 
> `/home/yosefk/Tools/BGP/Simulator/BGP/cbgp-2.0.0-rc3/src'
> make[2]: *** [check-am] Error 2
> make[2]: Leaving directory 
> `/home/yosefk/Tools/BGP/Simulator/BGP/cbgp-2.0.0-rc3/src'
> make[1]: *** [check-recursive] Error 1
> make[1]: Leaving directory 
> `/home/yosefk/Tools/BGP/Simulator/BGP/cbgp-2.0.0-rc3/src'
> make: *** [check-recursive] Error 1
>
> Yosef Kahana, Senior Software Engineer, Oversi
> M: +972-52-3477562 | T: +972-77-333-7771 | F: +972-3-542-3165
>
> yo...@ov...  |www.oversi.com


-- 
Prof. Bruno Quoitin                         Networking Lab
UMons, Belgium    http://informatique.umons.ac.be/networks
Phone: ++32 65 37 34 48             GSM: ++32 498 28 12 21

Re: [C-bgp-users] Q: C-BGP usage.

[Bruno Q. <bru...@um...>]

Dear Yosef,

Thank you for your interest. Please note that the latest (and most 
stable) version is 2.0.0-rc3 which is available from sourceforge at 
http://sourceforge.net/projects/c-bgp/
I also cc: the c-bgp mailing list on sourceforge.

> Remarks:
>
>    1. The documentation in the WEB is of version 1.1.20, lot of
>       command had been added/changed/removed. if there is an updated
>       User Guide, I'll be happy to have it.
>
The documentation is outdated. We are working towards a more recent user 
guide. Meanwhile, some of the newest commands are documented on-line at 
the following URL http://cbgp.info.ucl.ac.be/doc/html/index.html. There 
is also a tentative tutorial on-line et 
http://cbgp.info.ucl.ac.be/tutorial.php

>    1. All my questions are for implementing a simulated BGP networks
>       with several Domain. The simulated network should implement iBGP
>       as well as eBGP
>
You will find examples in the on-line tutorial. A good source for 
examples can also be found in the source archive under the "examples" 
directory. You will find examples for several scenari.

>
> Questions:
>
>    1. Topology file - I want to generate a topology file that has
>       several routers in the same AS. The example files describe each
>       router in a separated AS.
>       How do I set some routers in the same AS?
>
To do this, you will need to build routers by hand. To create a router 
involves several steps:

     net add node <addr>

This will create a node modeling the router: <addr> is the identifier of 
the node in the simulation. The identifier is formatted as an IP 
address, this address will automatically be allocated as a loopback 
address to the router, you can prevent this with the --no-loopback option)

If the node is going to belong to an IGP (interior gateway protocol), 
you can also create an IGP domain and add the node to this domain (this 
will be used to compute shortest path routes between all nodes in that 
domain).

     net add domain <id> igp
     net node <addr> domain <id>

Note that the first command is needed only once per domain. Once all 
your nodes (and the links between them have been setup and configured 
ith an IGP weight), you can run a single command to compute the shortest 
paths routes between the nodes within that domain:

     net domain <id> compute

Note that if you later change an IGP weight or change the status of an 
interface/link, you can call this command again to update the routes 
withint that domain.

Finally, you will need to create a BGP router in your node:

     bgp add router <asn> <addr>

The <asn> parameter is the AS number and <addr> is the identifier of an 
existing "node".

>    1. Create links - the documentation describe a "net node <address>
>       spf-prefix <prefix>" which no longer exist.
>       What is the new matching command?
>

The "spf-prefix" as used to computer shortest paths from one node 
towards all the nodes matching a given prefix. This has been replaced by 
the concept of "IGP domain" as described above.

To setup links, there are several options depending on the level of 
details of your model. The current c-bgp version is able to model 
router-to-router links, point-to-point links and LANs (we call them 
point-to-multipoint links).

An router-to-router link is the simplest way to connect nodes. It can be 
created with the following command where <addr1> and <addr2> identify 
existing nodes.

     net add link <addr1> <addr2>
     net add link 1.0.0.1 1.0.0.2

You cannot add parallel router-to-router links.

A point-to-point link is a bit more complex to setup. Network interfaces 
will be created on the nodes and the link will connect thoses interfaces 
together. Interface addresses (network / prefix_size) need to be 
specified for each end of the link. <addr1> and <addr2> are the node 
identifiers (same as above). <addr_if1> and <addr_if2> are the interface 
identifiers. In the following example, the network interface on node 
1.0.0.1 receives address 192.168.0.1 (prefix length is 30). while the 
other interface receives address 192.168.0.2.

     net add link-ptp <addr1> <addr_if2> <addr2> <addr_if2>
     net add link-ptp 1.0.0.1 192.168.0.1/30 1.0.0.2 192.168.0.2/30

Note that c-bgp currently checks that both interfaces are in the same 
subnet.

If you want to model a LAN, you first need to create the LAN, which is 
called a "subnet" in the c-bgp terminlogy. Then, you create links 
between your nodes and that LAN. A LAN is identified by a prefix and has 
a type (transit / stub) which is used by the IGP model to check if 
shortest paths can traverse that LAN (transit) or not (stub).

     net add subnet <prefix> <transit|subnet>
     net add link <addr> <addr_if>

     net add subnet 192.168.0.0/28
     net add link 1.0.0.1 192.168.0.1/28
     net add link 1.0.0.2 192.168.0.2/28
     net add link 1.0.0.3 192.168.0.7/28

Note that c-bgp currently checks interfaces on the same subnet have 
different addresses.

The above commands define "physical" links between nodes. You will also 
need to setup "BGP sessions" between BGP routers. This is done in a 
slightly different manner: suppose that you want to establish a BGP 
session between routers R1 and R2. You will need to setup the session on 
both routers (see example below). The "add peer" sub-command" defines a 
BGP neighbor on the router. The "peer up" command sets the session as 
administratively opened. The session will really be established if the 
two routers are reachable (you need to setup the underlying routes 
before, using an IGP and/or static routes).

     bgp router R1
         add peer R2
         peer R2 up

     bgp router R2 add peer R1
         add peer R1
         peer R1 up

The type of the session (iBGP or eBGP) is automatically derived from the 
router's ASN when the session is established.

>    1. Community - I would like to set communities at specific Domains.
>       can it be done at the domain boarder itself or should it be set
>       in the domain peer boarder?
>
It is done on a per-session basis, using filters. An example is shown 
below. The set of commands define a new rule in the outbound filter of 
router R1 for session with neighbor R2. The rules says that all routes 
(match any) will be added the community 100:2000

     bgp router R1
         peer R2
             filter out
                 add-rule
                     match any
                     action "community add 100:2000"



>    1. eBGP - distributing routing between Domains. The command net
>       options igp-inter not longer exist.
>       what is the configure eBGP updates distribution command?
>       How can we filter the updates of specific networks? we would
>       like to distribute some networks and block others.
>
Using filters (example above)

>    1. Router Reflector - we would like to set a router reflector in
>       the simulation.
>       How can it be done?
>

A router becomes a route reflector if it has BGP sessions with 
route-reflector clients. You can change a BGP session into a session 
with a client as follows

     bgp router R1 peer R2 rr-client

That makes automatically R1 a RR.

>    1. BGP listener node - we would like to set a node that is a
>       listener to the Router Reflector. We would like to record all
>       the updates deliver to this listener and to dump it to a file.
>       Please advise about this issue.
>

     bgp router R1 peer R2 record <file>


>    1. Thanks for the support.
>

Hope this helps

Please tell me if you need additional support (We have in-house scripts 
and applications that allow to more easily setup c-bgp models using real 
network data).

And finally, I would appreciate if you can describe a bit more what's 
your exact application ?

Regards,
Bruno







-- 
Prof. Bruno Quoitin                         Networking Lab
UMons, Belgium    http://informatique.umons.ac.be/networks
Phone: ++32 65 37 34 48             GSM: ++32 498 28 12 21

Re: [C-bgp-users] Filters in cbgp-2.0.0 rc

[Bruno Q. <bru...@um...>]

*Dear Palani,*

The syntax has slightly changed since version 1.4. Unfortunately, there 
is no up-to-date documentation. However, have a look at file 
"examples/example-net-export.cli" in the c-bgp source archive. It uses 
BGP filters extensively and is a good start to discover the filter 
syntax. In case of doubt, please contact me again.

Regards,
Bruno
*

[C-bgp-users] Filters in cbgp-2.0.0 rc 
<https://sourceforge.net/mailarchive/message.php?msg_id=26759366>*
Delete 
<https://sourceforge.net/mailarchive/exclude_list.php?action=addexclude&group_id=276050&forum_name=c-bgp-users&msg_name=AANLkTim7QsF%2BhzrsiPGJm-a%3D-%2BEi2wdAvLDO2F3UqcE2%40mail.gmail.com> 
From: Palani kodeswaran <palani.k@gm...> - 2010-12-09 00:07

*Attachments:* Message as HTML 
<https://sourceforge.net/mailarchive/attachment.php?list_name=c-bgp-users&message_id=AANLkTim7QsF%2BhzrsiPGJm-a%3D-%2BEi2wdAvLDO2F3UqcE2%40mail.gmail.com&counter=1> 


Hello,

Is the filter add-rule syntax the same in CBGP 1.4 and CBGP 2. Also, where
can I find documentation on filters for CBGP 2.

Thanks,
Palani.



-- 
Prof. Bruno Quoitin                         Networking Lab
UMons, Belgium    http://informatique.umons.ac.be/networks
Phone: ++32 65 37 34 48             GSM: ++32 498 28 12 21

[C-bgp-users] Filters in cbgp-2.0.0 rc

[Palani k. <pal...@gm...>]

Hello,

Is the filter add-rule syntax the same in CBGP 1.4 and CBGP 2. Also, where
can I find documentation on filters for CBGP 2.

Thanks,
Palani.

[C-bgp-users] Filters in cbgp-2.0.0 rc

[Palani k. <pal...@gm...>]

Hello,

Is the filter add-rule syntax the same in CBGP 1.4 and CBGP 2. Also, where
can I find documentation on filters for CBGP 2.

Thanks,
Palani.

[C-bgp-users] Fwd: [cbgp] Node addresses from AS numbers while using bgp load topology

[Bruno Q. <bru...@um...>]

Dear Palani,

[The new mailing-list is now on sourceforge]

The conversion depends on the c-bgp version you are using. In the past, 
it was a fixed conversion were the ASN was used in the two most 
significant bytes of the IP address, the two least significant bytes 
were 0. For example, ASN7018 would have a single router with address 
27.106.0.0 (since 7018=27*256 + 106).

With c-bgp 2.0.0 the ASN <->IP address mapping can be configured through 
so-called addressing schemes. The default behaviour is the same as past 
version however. The addressing scheme can be mentionned as an option to 
the "bgp topology load" command. Only two schemes are possible today: 
"default" (7018 -> 27.106.0.0) and "local" (7018 -> 0.0.27.106). The 
second one was introduced to better control where router addresses were 
located (always in prefix 0.0/16) and avoid routing perturbations where, 
for example, a BGP prefix advertisement would change the way two routers 
can reach each other.

Cheers,
Bruno

-------- Original Message --------
Subject: 	[cbgp] Node addresses from AS numbers while using bgp load 
topology
Date: 	Mon, 29 Nov 2010 02:34:04 -0500
From: 	Palani kodeswaran <pal...@gm...>
To: 	cbg...@li...



Hello,

I am trying to understand how AS numbers in a topology file used in "bgp 
topology file" are assigned IP addresses. For example, what ip would AS 
7018 map to.

Thanks,
Palani.