[B.A.T.M.A.N.] Batman, TTL, broadcast, etc. (UNCLASSIFIED)

[B.A.T.M.A.N.] Batman, TTL, broadcast, etc. (UNCLASSIFIED)

[Karan, Cem F CIV (US)]

[-- Attachment #1: Type: text/plain, Size: 1647 bytes --]

Classification: UNCLASSIFIED
Caveats: NONE

Hi all, I'm trying to get some information on how batman works to determine
what my next steps are for an experiment I'm running.  I'm hoping someone on
the list can tell me how Batman operates RIGHT NOW, not just how it is
intended to operate in the future, so I can make reasonable plans. 

1) I'm guessing from the following pages (which only describe roaming and
announcement behaviors) that the TTL field of all packets is decremented by
1.  Is this true?
http://www.open-mesh.org/projects/batman-adv/wiki/Client-roaming
http://www.open-mesh.org/projects/batman-adv/wiki/Client-announcement

2) Follow up to question 1; are the TTL fields of Batman packets and IP
packets linked in some way?  The library I'm using (zeromq,
http://www.zeromq.org/) has a reliable multicast transport built on top of
OpenPGM (http://code.google.com/p/openpgm/).  My plan is to simulate
reliable 1-hop broadcast by using reliable multicast and setting the TTL
field to 1.  However, this will only affect the IP layer.  Unless batman
also decrements the TTL field of the IP packets traveling over it, I'm kind
of stuck.

3) Finally, does batman have the equivalent of multicast or (better yet)
broadcast for data packets?  That is, if I send something to a multicast IP
address which all of its 1-hop neighbors are listening to, will all of them
listen to the packet simultaneously, or will it act like a series of unicast
messages?

If you're wondering what I'm trying to do, message me directly.  It's easy
to explain, but off topic for this list.

Thanks,
Cem Karan   

Classification: UNCLASSIFIED
Caveats: NONE



[-- Attachment #2: smime.p7s --]
[-- Type: application/x-pkcs7-signature, Size: 5559 bytes --]

Re: [B.A.T.M.A.N.] Batman, TTL, broadcast, etc. (UNCLASSIFIED)

[Sven Eckelmann]

[-- Attachment #1: Type: text/plain, Size: 2070 bytes --]

On Tuesday 04 September 2012 12:48:00 Karan, Cem F CIV wrote:
[...]
> 1) I'm guessing from the following pages (which only describe roaming and
> announcement behaviors) that the TTL field of all packets is decremented by
> 1.  Is this true?
> http://www.open-mesh.org/projects/batman-adv/wiki/Client-roaming
> http://www.open-mesh.org/projects/batman-adv/wiki/Client-announcement

TTL fields of the batman-adv specific TTL fields are decremented by one for 
each hop.

> 2) Follow up to question 1; are the TTL fields of Batman packets and IP
> packets linked in some way?  The library I'm using (zeromq,
> http://www.zeromq.org/) has a reliable multicast transport built on top of
> OpenPGM (http://code.google.com/p/openpgm/).  My plan is to simulate
> reliable 1-hop broadcast by using reliable multicast and setting the TTL
> field to 1.  However, this will only affect the IP layer.  Unless batman
> also decrements the TTL field of the IP packets traveling over it, I'm kind
> of stuck.

No, batman-adv is a distributed switch on layer 2. So it operates on a 
different layer and has no knowledge of IP or its TTL field (please ignore the 
gateway feature for now... this is the only exception were it accesses L3 
stuff).

> 3) Finally, does batman have the equivalent of multicast or (better yet)
> broadcast for data packets?  That is, if I send something to a multicast IP
> address which all of its 1-hop neighbors are listening to, will all of them
> listen to the packet simultaneously, or will it act like a series of unicast
> messages?

It currently has layer 2 broadcast support (it resents broadcast packets 3 
times to increase the propability for a successful tx). Multicast is handled 
as multicast. Linus Luessing has a patchset to implement optimized multicast 
[1]. I am not sure about the current state. But maybe he can give more 
detailed information about it. But you should know that the optimized 
implementation may use unicast in some scenarios.

Kind regards,
	Sven

[1] http://www.open-mesh.org/projects/batman-adv/wiki/Multicast-ideas

[-- Attachment #2: This is a digitally signed message part. --]
[-- Type: application/pgp-signature, Size: 836 bytes --]

Re: [B.A.T.M.A.N.] Batman, TTL, broadcast, etc. (UNCLASSIFIED)

[Karan, Cem F CIV (US)]

[-- Attachment #1: Type: text/plain, Size: 3358 bytes --]

Classification: UNCLASSIFIED
Caveats: NONE

OK, based on the answers below, I have a few more questions.

1) Is there any way for a user program to communicate with batman if certain
packets need special handling?  I don't mind if I have to mark each packet
individually, or if I have to do something else interesting, but if I can
mark a packet as special, then I can hack code into batman to set the TTL
field the way I need it.

2) Does batman rely on multicast in any way for its own operation?  That is,
for something that keeps the network running well.  If not, I can force all
multicast packets to have a TTL of 1.  This still requires some way of
marking my own IP packets as multicast packets in a manner that batman can
recognize, but I'm willing to do that, if it will help.

Thanks,
Cem Karan

> -----Original Message-----
> From: Sven Eckelmann [mailto:sven@narfation.org]
> Sent: Tuesday, September 04, 2012 9:05 AM
> To: b.a.t.m.a.n@lists.open-mesh.org
> Cc: Karan, Cem F CIV (US); Linus Lüssing
> Subject: Re: [B.A.T.M.A.N.] Batman, TTL, broadcast, etc. (UNCLASSIFIED)
> 
> On Tuesday 04 September 2012 12:48:00 Karan, Cem F CIV wrote:
> [...]
> > 1) I'm guessing from the following pages (which only describe roaming
> > and announcement behaviors) that the TTL field of all packets is
> > decremented by 1.  Is this true?
> > http://www.open-mesh.org/projects/batman-adv/wiki/Client-roaming
> > http://www.open-mesh.org/projects/batman-adv/wiki/Client-announcement
> 
> TTL fields of the batman-adv specific TTL fields are decremented by one
> for each hop.
> 
> > 2) Follow up to question 1; are the TTL fields of Batman packets and
> > IP packets linked in some way?  The library I'm using (zeromq,
> > http://www.zeromq.org/) has a reliable multicast transport built on
> > top of OpenPGM (http://code.google.com/p/openpgm/).  My plan is to
> > simulate reliable 1-hop broadcast by using reliable multicast and
> > setting the TTL field to 1.  However, this will only affect the IP
> > layer.  Unless batman also decrements the TTL field of the IP packets
> > traveling over it, I'm kind of stuck.
> 
> No, batman-adv is a distributed switch on layer 2. So it operates on a
> different layer and has no knowledge of IP or its TTL field (please
> ignore the gateway feature for now... this is the only exception were
> it accesses L3 stuff).
> 
> > 3) Finally, does batman have the equivalent of multicast or (better
> > yet) broadcast for data packets?  That is, if I send something to a
> > multicast IP address which all of its 1-hop neighbors are listening
> > to, will all of them listen to the packet simultaneously, or will it
> > act like a series of unicast messages?
> 
> It currently has layer 2 broadcast support (it resents broadcast
> packets 3 times to increase the propability for a successful tx).
> Multicast is handled as multicast. Linus Luessing has a patchset to
> implement optimized multicast [1]. I am not sure about the current
> state. But maybe he can give more detailed information about it. But
> you should know that the optimized implementation may use unicast in
> some scenarios.
> 
> Kind regards,
> 	Sven
> 
> [1] http://www.open-mesh.org/projects/batman-adv/wiki/Multicast-ideas

Classification: UNCLASSIFIED
Caveats: NONE



[-- Attachment #2: smime.p7s --]
[-- Type: application/x-pkcs7-signature, Size: 5559 bytes --]

Re: [B.A.T.M.A.N.] Batman, TTL, broadcast, etc. (UNCLASSIFIED)

[Sven Eckelmann]

[-- Attachment #1: Type: text/plain, Size: 1484 bytes --]

On Wednesday 05 September 2012 13:32:29 you wrote:
> Classification: UNCLASSIFIED
> Caveats: NONE
> 
> OK, based on the answers below, I have a few more questions.
> 
> 1) Is there any way for a user program to communicate with batman if certain
> packets need special handling?  I don't mind if I have to mark each packet
> individually, or if I have to do something else interesting, but if I can
> mark a packet as special, then I can hack code into batman to set the TTL
> field the way I need it.

How would you signal other layers you don't interact with directly? There is 
no special socket API that allows you do modify the batman-adv stuff. But you 
can for example do things like the gateway handling which checks different 
layer 3/4 things and modifies the behaviour slightly.

So it is (depending on the problem) doable, but there is no general solution 
available.

> 2) Does batman rely on multicast in any way for its own operation?  That is,
> for something that keeps the network running well.  If not, I can force all
> multicast packets to have a TTL of 1.  This still requires some way of
> marking my own IP packets as multicast packets in a manner that batman can
> recognize, but I'm willing to do that, if it will help.

Multicast on ethernet... yes. Because it does broadcasts to send OGMs. But on 
IP level - no, batman-adv doesn't work on top of layer 3.

And marking the packets. This could be done using special multicast addresses.

Kind regards,
	Sven

[-- Attachment #2: This is a digitally signed message part. --]
[-- Type: application/pgp-signature, Size: 836 bytes --]

Re: [B.A.T.M.A.N.] Batman, TTL, broadcast, etc. (UNCLASSIFIED)

[Karan, Cem F CIV (US)]

[-- Attachment #1: Type: text/plain, Size: 1876 bytes --]

On 9/8/12 11:37 AM, "Sven Eckelmann" <sven@narfation.org> wrote:

>On Wednesday 05 September 2012 13:32:29 you wrote:
>> Classification: UNCLASSIFIED
>> Caveats: NONE
>> 
>> OK, based on the answers below, I have a few more questions.
>> 
>> 1) Is there any way for a user program to communicate with batman if
>>certain
>> packets need special handling?  I don't mind if I have to mark each
>>packet
>> individually, or if I have to do something else interesting, but if I
>>can
>> mark a packet as special, then I can hack code into batman to set the
>>TTL
>> field the way I need it.
>
>How would you signal other layers you don't interact with directly? There
>is 
>no special socket API that allows you do modify the batman-adv stuff. But
>you 
>can for example do things like the gateway handling which checks
>different 
>layer 3/4 things and modifies the behaviour slightly.
>
>So it is (depending on the problem) doable, but there is no general
>solution 
>Available.

That was what I was afraid of.  I'll have to do something else then.

>> 2) Does batman rely on multicast in any way for its own operation?
>>That is,
>> for something that keeps the network running well.  If not, I can force
>>all
>> multicast packets to have a TTL of 1.  This still requires some way of
>> marking my own IP packets as multicast packets in a manner that batman
>>can
>> recognize, but I'm willing to do that, if it will help.
>
>Multicast on ethernet... yes. Because it does broadcasts to send OGMs.
>But on 
>IP level - no, batman-adv doesn't work on top of layer 3.
>
>And marking the packets. This could be done using special multicast
>addresses.

That might work for me!  I have few enough nodes that I can assign
numerous special multicast addresses to each node, for each purpose I have
in mind.  I'll have to work on it.  Thank you for the suggestion!

Thanks,
Cem Karan

[-- Attachment #2: smime.p7s --]
[-- Type: application/pkcs7-signature, Size: 5281 bytes --]