[OLSR-users] ICMP Redirects

Sven-Ola Tuecke (spam-protected)
Wed Aug 31 11:51:00 CEST 2005


Lolo,

well, ICMP redirects are contraproductive. Besides the fact that the packets 
will waste bandwidth, no OLSR node should reconfigure it's routing table in 
reaction of ICMP-Redir. Why? The anncounced next hop may be unreachable even 
if configured for the same subnet. So simply switch them off when using 
OLSR.

OK - the officials may state: all IPs matching a netmask should be reachable 
because they form a broadcast domain. OLSR depends on broadcasting and the 
255.255.255.255 netmask ist for non-broacasting-interfaces (e.g. modem link) 
only.

In practice, the only meaningful use of netmask 255.255.255.255 is: A device 
with 2 interfaces where you do not want to waste any IPs for a 
broadcast/network address. Example: We often have this config in Berlin:

Node1:
  WLAN0: 104.0.2.9 / 255.255.255.252 (Master)
  WLAN1: 104.111.222.333 / 255.0.0.0 (Ad-Hoc)
Node2:
  WLAN0: 104.0.2.10 / 255.255.255.252 (Managed)
  WLAN1: 104.111.222.334 / 255.0.0.0 (Ad-Hoc)

In this sample, OLSR fires both on the managed link (for longer distances) 
and on the Ad-Hoc mesh interface (for the neighbourhood). Here 104.0.2.8 
(==network) and 104.0.2.11 (==broadcast) are wasted IP addresses for the 
"mini-network".

For a more IP-economical setup, the netmask 255.255.255.255 in conjunction 
with any other broadcast address is usable with Linux. Traffic is routed 
through this iface by host routes only. Example: "ifconfig WLAN0 104.0.2.9 
netmask 255.255.255.255 broadcast 104.255.255.255" will do the job here too. 
And as you already pointed out: XP does not like the 255.255.255.255 
either...

LG, Sven-Ola

"lolo" <(spam-protected)> schrieb im Newsbeitrag 
news:(spam-protected)
> Hi every body
>
> Weber, Michael J. (US SSA) wrote:
>
>> Under Win32, there is a call made to disable redirects. Apparently,
>> linux will issue a redirect if it forwards a packet out the same
>> interface on which it was received (which is always the case with a
>> transit node in a MANET.)
>
> There are some more subtilities about ICMP redirect. From the network
> sorcery web site :
> http://www.networksorcery.com/enp/protocol/icmp/msg5.htm
>
>> Routers MUST NOT generate a Redirect Message unless *all* the
>> following conditions are met:
>>
>> * The packet is being forwarded out the same physical interface
>> that it was received from.
>> * The IP source address in the packet is on the same Logical IP
>> (sub)network as the next-hop IP address.
>> * The packet does not contain an IP source route option.
>>
>> The source address used in the ICMP Redirect MUST belong to the
>> same logical (sub)net as the destination address.
>>
>> A router using a routing protocol (other than static routes) MUST
>> NOT consider paths learned from ICMP Redirects when forwarding a
>> packet. If a router is not using a routing protocol, a router MAY
>> have a configuration that, if set, allows the router to consider
>> routes learned through ICMP Redirects when forwarding packets.
>>
>
> "Regular" configuration of MANET nodes meets these three conditions
>
> So I understand why at the 2nd OLSR Interop this summer, the guys from
> INRIA and LRI insist loudly on configuring our interface with a netmask of
> 255.255.255.255, even we were all using networks of the form 10.x.0.0/8
> (we were ask to choose a value for x in the [1,9] for identification of 
> the
> teams nodes).
>
> That Netmask could be translated as : i am the only node reachable on
> that network, which is my IP address (weird), so don't bother me with
> icmp redirect, and don't even try to send that IP frame directly to the
> network but use instead a gateway in every cases.
>
> BTW it was not being possible to set that value on windows XP
> (although with found a possible hack at editing the registry at the end of
> the day but with no testing)
>
> My two cents
>
> Laurent
>
>
> _______________________________________________
> olsr-users mailing list
> (spam-protected)
> https://www.olsr.org/mailman/listinfo/olsr-users 




More information about the Olsr-users mailing list