[Click] DSR+ETX DSRArpTable problems

[Florian Sesser]

Hi!

As the grid-hackers list does not exist anymore, I'll allow myself to
repost my inquery on this list.

I am having issues getting DSR+ETX to run well.

Thanks,

Florian

-------- Original Message --------
Subject: [Bowl-users] DSR+ETX DSRArpTable problems
Date: Thu, 25 Nov 2010 05:07:46 +0100
From: Florian Sesser <sesser at in.tum.de>
To: grid-hackers at pdos.lcs.mit.edu, 	BOWL Users
<bowl-users at lists.net.t-labs.tu-berlin.de>

Hi!

Please excuse my cross-posting this inquiry to grid-hackers as well as
bowl-users.

I'm having a hard time getting DSR+ETX to run properly. I have only once
gotten a route with more than 2 hops to work, in a testbed of ~ 35
nodes. OLSR and (even) our own SSR perform much better (I get links with
6 hops and more). It worked for a few seconds (like in 20 out of 600
pings came through on a four-hop path).

It looks like DSRArpTable does not work as it should.

I get lots of these:

> ma-roof-1-avila: DSRArpTable::push:  missing ARP table entry!  src: 10.20.5.9/00-0B-6B-84-B2-3C dst: 10.20.5.9/00-00-00-00-00-00
(note: ma-roof-1-avila _is_ 10.20.5.9! also, its MAC really is
00:0B:6B:84:B2:3C)
Also, the DSRArpTable shouldn't get any packets from the local host.

and some of these:
> af-roof-3-avila: DSRArpTable::push:  existing entry for 10.20.9.9 has different MAC!  00-0B-6B-84-B3-5F, not 00-0B-6B-84-B0-26
> af-roof-3-avila: DSRArpTable::push:  existing entry for 10.20.9.9 has different MAC!  00-0B-6B-84-B0-26, not 00-0B-6B-84-B3-5F
> af-roof-4-avila: DSRArpTable::push:  missing ARP table entry!  src: 10.20.4.33/00-0B-6B-84-B3-5F dst: 10.20.12.25/00-00-00-00-00-00
> af-roof-3-avila: DSRArpTable::push:  existing entry for 10.20.9.9 has different MAC!  00-0B-6B-84-B3-5F, not 00-0B-6B-84-B0-26
> af-roof-3-avila: DSRArpTable::push:  existing entry for 10.20.9.9 has different MAC!  00-0B-6B-84-B0-26, not 00-0B-6B-84-B3-5F
> af-roof-4-avila: DSRArpTable::push:  missing ARP table entry!  src: 10.20.4.33/00-0B-6B-84-B3-5F dst: 10.20.12.25/00-00-00-00-00-00
> af-roof-4-avila: DSRArpTable::push:  missing ARP table entry!  src: 10.20.4.33/00-0B-6B-84-B3-5F dst: 10.20.12.25/00-00-00-00-00-00

af-roof-3-avila (10.10.4.25) is right only half of the time:
The MAC of the mesh interface of 10.20.9.9 is 00:0B:6B:84:B0:26

For the sake of completeness, I'll attach my whole Click configuration.
The most interesting part probably is:

> FromDevice(mesh0, SNIFFER false, PROMISC false)
>         -> HostEtherFilter($myaddr_ethernet,1)
>         -> class;

> class :: Classifier(12/0806 20/0001, 12/0806 20/0002, 12/7ffe, 12/7ffd, -);

> class[4]
>         -> CheckIPHeader(14)
>         -> [2]dsr_arp[2]
>         -> Strip(14)
>         -> GetIPAddress(16)
>         -> DecIPTTL // Flo: IP TTL == DSR TTL.
>         -> DSR_class1 :: Classifier(09/C8,-); //Check for DSR packets 

> DSR_class1[0] -> [1]dsr_rt;

Nothing special so far.
It's all pretty much like what Click's conf/make-dsr-config.pl emits,
adopted to BOWL and with some small changes.

If you have any hints for me (or can tell me that this is normal for
this implementation) I'd be most thankful.

Regards,

Florian

-------------- next part --------------
An embedded and charset-unspecified text was scrubbed...
Name: etx_dsr_with_ipmp.cl
Url: http://amsterdam.lcs.mit.edu/pipermail/click/attachments/20101125/5e41ce13/attachment.txt 
-------------- next part --------------
An embedded and charset-unspecified text was scrubbed...
Name: Attached Message Part
Url: http://amsterdam.lcs.mit.edu/pipermail/click/attachments/20101125/5e41ce13/attachment-0001.txt