[B.A.T.M.A.N.] 0.3 alpha (today revison)

[B.A.T.M.A.N.] 0.3 alpha (today revison)

[a.anselmi]

Hi,
since you use mipsel device I think these mips log could be of more help
in fixing mips/mipsel bug in 0.3 branches (alpha and experimental).

Very simple scenario:
[client 5.1.7.251]----mesh----[gateway 5.1.7.254]
all node running 0.3 apha (the today latest rv6300.3-alpha)

##### d3 level debug at client
...
1) Adding throw route to 169.254.0.0/16 via 0.0.0.0 (table 68 - eth0:1)
2) Adding throw route to 127.0.0.0/8 via 0.0.0.0 (table 68 - lo)
3) Adding throw route to 10.7.251.128/25 via 0.0.0.0 (table 68 - ath2)
4) Adding throw route to 5.0.0.0/8 via 0.0.0.0 (table 68 - ath0)
5) Adding throw route to 10.7.251.0/25 via 0.0.0.0 (table 68 - br0)
6) Adding route to 5.1.7.254 via 0.0.0.0 (table 66 - ath0)
7) Found new gateway 5.1.7.254 -> class: 49 - 4MBit/1024KBit

# after sliding windows if filled in by validate OGMs

8) Adding default route to 5.1.7.254 (gw_flags: 49, packet_count: 64,
gw_product: 0)
9) Adding default route via tun0 (table 68)

# ...ah this is great! now trying to ping outside mesh

10) Gateway client - got IP (1.0.254.169) from gateway: 5.1.7.254
11) Gateway client - gateway (5.1.7.254) says: IP (1.0.254.169) is expired

# client can't exit off mesh

##### d3 level debug at gateway
...
1) gateway class: 49 -> propagating: 4MBit/1024KBit
2) Error - can't add route to 0.0.254.169/24 via 0.0.0.0 (table 254):
Invalid argument

# (don't know why previous error)
# ...gateway is learned about client wants ping outside mesh

3) Gateway - assigned 1.0.254.169 to client 5.1.7.251
4) Error - got packet from unknown client: 5.1.7.251 (virtual ip 1.0.254.169)

What I see:

FIRST BUG (?)
Client prepare route to 169.254.0.0/16 in table 68 (line 1-log_client)
while gateway arrange a 169.254.0.0/24 space (line 2-log_gateway): suffix
bits differs (16 bit netmask client side vs 24 bit netmask server side)

SECOND BUG
Client wants ping outside mesh and receives a *little endian formatted* IP
1.0.254.169 (line 10-log_client and line 3-log_gateway) which is out of
previous arrangement (this IP address is malformed by gateway).

Client tryes sending outgoing packet to gateway

Gateway receives client' outgoing packect but doesn't recognize IP source
address (line 4-log_gateway) expecting one in 0.0.254.169/24 space. Thus
gateway informs client about wrong IP, so client pinging fails (line
11-log_client).

Hopeing this will be usefull... I know today weekend begins :)


--
Antonio

Re: [B.A.T.M.A.N.] 0.3 alpha (today revison)

[Marek Lindner]

Hi,

> since you use mipsel device I think these mips log could be of more help
> in fixing mips/mipsel bug in 0.3 branches (alpha and experimental).

> Hopeing this will be usefull... I know today weekend begins :)

thanks a lot for your thorough description - it did help.  :-)
Since the weekend is shorter than one might think I already issued a patch. It 
should work now but due to the lack of Big Endian CPUs I can't test it. I 
would be happy if you checked on that. After you confirmed the end of this 
bug I will fix the experimental branch as well.

Regards,
Marek

Re: [B.A.T.M.A.N.] 0.3 alpha (today revison)

[a.anselmi]

great work Marek,
now client can ping outside mesh obtaining the correct IP from gateway.
Thanks!

Still remain the minor problem in arrangement of IPIPA subnet mask:
169.254.0.0/16 on clients
169.254.0.0/24 on gateway

Have a fine weekend


--
Antonio


space (line 2-log_gateway): suffix
> Hi,
>
>> since you use mipsel device I think these mips log could be of more help
>> in fixing mips/mipsel bug in 0.3 branches (alpha and experimental).
>
>> Hopeing this will be usefull... I know today weekend begins :)
>
> thanks a lot for your thorough description - it did help.  :-)
> Since the weekend is shorter than one might think I already issued a
> patch. It
> should work now but due to the lack of Big Endian CPUs I can't test it. I
> would be happy if you checked on that. After you confirmed the end of this
> bug I will fix the experimental branch as well.
>
> Regards,
> Marek
> _______________________________________________
> B.A.T.M.A.N mailing list
> B.A.T.M.A.N@open-mesh.net
> https://list.open-mesh.net/mm/listinfo/b.a.t.m.a.n
>

Re: [B.A.T.M.A.N.] 0.3 alpha (today revison)

[Marek Lindner]

Am Samstag, 22. September 2007 20:00:04 schrieb a.anselmi@oltrelinux.com:
> great work Marek,
> now client can ping outside mesh obtaining the correct IP from gateway.
> Thanks!

Goooood. This makes the weekend even better.  ;-)


> Still remain the minor problem in arrangement of IPIPA subnet mask:
> 169.254.0.0/16 on clients
> 169.254.0.0/24 on gateway

I don't know where you can see that. The tun interface is a point to point 
connection ...


Greetings,
Marek


PS: 2 hours reponse time isn't that bad, is it ?  :)

Re: [B.A.T.M.A.N.] 0.3 alpha (today revison)

[a.anselmi]

>> Still remain the minor problem in arrangement of IPIPA subnet mask:
>> 169.254.0.0/16 on clients
>> 169.254.0.0/24 on gateway
>
> I don't know where you can see that. The tun interface is a point to > >
> point connection ...
...sure, it was a safe-IP on my router which cause that messages about
169.254.0.0 route: now changed safe-IP out of ipipa subnet.

> PS: 2 hours reponse time isn't that bad, is it ?  :)

...that's the way! (K.C. & Sunshine Band, early 70's)

ciao

--
Antonio

Re: [B.A.T.M.A.N.] 0.3 alpha (today revison)

[Marek Lindner]

Am Samstag, 22. September 2007 21:41:21 schrieb a.anselmi@oltrelinux.com:
> >> Still remain the minor problem in arrangement of IPIPA subnet mask:
> >> 169.254.0.0/16 on clients
> >> 169.254.0.0/24 on gateway
> >
> > I don't know where you can see that. The tun interface is a point to > >
> > point connection ...
>
> ...sure, it was a safe-IP on my router which cause that messages about
> 169.254.0.0 route: now changed safe-IP out of ipipa subnet.

Excuse my ignorance but what is "ipipa" ? Even wikipedia does not know 
it.  ;-)
And let me repeat my question: Where did you see that routing 
entry "169.254.0.0/16" ? Or is that problem also gone ?


Greetings,
Marek

Re: [B.A.T.M.A.N.] 0.3 alpha (today revison)

[Steven Leeman]

[-- Attachment #1: Type: text/plain, Size: 1078 bytes --]

he means Apipa...
http://compnetworking.about.com/cs/protocolsdhcp/g/bldef_apipa.htm
as soon as something starts with 169.x.y.z I guess.. netmask 255.0.0.0


2007/9/22, Marek Lindner <lindner_marek@yahoo.de>:
>
> Am Samstag, 22. September 2007 21:41:21 schrieb a.anselmi@oltrelinux.com:
> > >> Still remain the minor problem in arrangement of IPIPA subnet mask:
> > >> 169.254.0.0/16 on clients
> > >> 169.254.0.0/24 on gateway
> > >
> > > I don't know where you can see that. The tun interface is a point to >
> >
> > > point connection ...
> >
> > ...sure, it was a safe-IP on my router which cause that messages about
> > 169.254.0.0 route: now changed safe-IP out of ipipa subnet.
>
> Excuse my ignorance but what is "ipipa" ? Even wikipedia does not know
> it.  ;-)
> And let me repeat my question: Where did you see that routing
> entry "169.254.0.0/16" ? Or is that problem also gone ?
>
>
> Greetings,
> Marek
> _______________________________________________
> B.A.T.M.A.N mailing list
> B.A.T.M.A.N@open-mesh.net
> https://list.open-mesh.net/mm/listinfo/b.a.t.m.a.n
>

[-- Attachment #2: Type: text/html, Size: 1866 bytes --]

Re: [B.A.T.M.A.N.] 0.3 alpha (today revison)

[elektra]

169.254.0.0/16 is the MDNS (multicast DNS) address space. Part of zeroconf 
repectively 'Bonjour' (Apple Computers)

This has nothing to do with Batman.

cu elektra

Re: [B.A.T.M.A.N.] 0.3 alpha (today revison)

[a.anselmi]

Hi,
ipipa was a my typo... I was meaning apipa.

"Windows 98, 98 SE, Me, and 2000 have an Automatic Private IP Addressing
(APIPA) feature that will automatically assign an Internet Protocol
address to a computer on which it installed.  This occurs when the TCP/IP
protocol is installed, set to obtain it's IP address automatically from a
Dynamic Host Configuration Protocol server,  and when there is no DHCP
server present or the DHCP server is not available.  The Internet Assigned
Numbers Authority (IANA) has reserved private IP addresses in the range of
169.254.0.0 -169.254.255.255 for Automatic Private IP Addressing."

good weekend to all batman team

Antonio



> 169.254.0.0/16 is the MDNS (multicast DNS) address space. Part of zeroconf
> repectively 'Bonjour' (Apple Computers)
>
> This has nothing to do with Batman.
>
> cu elektra
> _______________________________________________
> B.A.T.M.A.N mailing list
> B.A.T.M.A.N@open-mesh.net
> https://list.open-mesh.net/mm/listinfo/b.a.t.m.a.n
>

RE: [B.A.T.M.A.N.] 0.3 alpha (today revison)

[Michael Burmeister-Brown]

Marek,

I hate to bother you but you were so responsive before...

Antonio just put up a post.  We are both seeing the same problem -
computers associated with repeaters on 0.3 (MIPS) can't browse.  But ssh
into that node and do a wget and it works great!.  If we just replace
0.3 with 0.2 all works again.

Also, I am tracking memory and see it dropping steadily on the nodes
(500k in about 12 hours).  This didn't happen with 0.2 and that is all
that has changed.

Is there any way you can look into this ASAP?  We are very close to
having to give code to our supplier for the routers and we would very
much like to ship with 0.3 rather than 0.2.

Thank you for your help!

FYI, we sent you 6 merakis today.  I'll get you a tracking number
shortly.

Mike

-----Original Message-----
From: b.a.t.m.a.n-bounces@open-mesh.net
[mailto:b.a.t.m.a.n-bounces@open-mesh.net] On Behalf Of Marek Lindner
Sent: Saturday, September 22, 2007 11:52 AM
To: The list for a Better Approach To Mobile Ad-hoc Networking
Subject: Re: [B.A.T.M.A.N.] 0.3 alpha (today revison)

Am Samstag, 22. September 2007 20:00:04 schrieb
a.anselmi@oltrelinux.com:
> great work Marek,
> now client can ping outside mesh obtaining the correct IP from
gateway.
> Thanks!

Goooood. This makes the weekend even better.  ;-)


> Still remain the minor problem in arrangement of IPIPA subnet mask:
> 169.254.0.0/16 on clients
> 169.254.0.0/24 on gateway

I don't know where you can see that. The tun interface is a point to
point 
connection ...


Greetings,
Marek


PS: 2 hours reponse time isn't that bad, is it ?  :)
_______________________________________________
B.A.T.M.A.N mailing list
B.A.T.M.A.N@open-mesh.net
https://list.open-mesh.net/mm/listinfo/b.a.t.m.a.n

Re: [B.A.T.M.A.N.] 0.3 alpha (today revison)

[Axel Neumann]

Hello,

On Montag 24 September 2007, Michael Burmeister-Brown wrote:
> Marek,
>
> I hate to bother you but you were so responsive before...
>
> Antonio just put up a post.  We are both seeing the same problem -
> computers associated with repeaters on 0.3 (MIPS) can't browse.  But ssh
> into that node and do a wget and it works great!.  If we just replace
> 0.3 with 0.2 all works again.

So you want to use the batman-clients as APs for usual 802.11-managed-mode 
clients.
If I understand you right, then you prbably need to do network address 
translation (NAT) or MASQUERADEing on the bat0 interfaces on the batman 
clients.
And of course a second time on the batman-GW.

>
> Also, I am tracking memory and see it dropping steadily on the nodes
> (500k in about 12 hours).  This didn't happen with 0.2 and that is all
> that has changed.
Thats interesting, sounds like a memory leak, because in the lates setups i 
tested the memory consumption was constantly 3.1 percent (even after hours). 
using dynamically compiled mipsel binaries and up to 20 nodes .
Can you send the output from top right after the start, after a few minutes 
and after a few hours. And the output of debug-level -5 (hope its compiled 
in)

ciao,
axel

Re: [B.A.T.M.A.N.] 0.3 alpha (today revison)

[Marek Lindner]

Hi,

> I hate to bother you but you were so responsive before...

no problem - I'm here.  :-)


> Antonio just put up a post.  We are both seeing the same problem -
> computers associated with repeaters on 0.3 (MIPS) can't browse.  But ssh
> into that node and do a wget and it works great!.  If we just replace
> 0.3 with 0.2 all works again.

From the client associated to AP_n you can ssh to your gateway ? Than you have 
a problem with your internet tunnel.


> Also, I am tracking memory and see it dropping steadily on the nodes
> (500k in about 12 hours).  This didn't happen with 0.2 and that is all
> that has changed.

What do you mean with "memory and see it dropping steadily" ?


> Is there any way you can look into this ASAP?  We are very close to
> having to give code to our supplier for the routers and we would very
> much like to ship with 0.3 rather than 0.2.

And you are sure to preinstall 0.3 ? It is not stable yet. There are a bunch 
of things to come soon: improved blackhole detection, BATMAN IV, a kernel 
modul for faster tunneling, bugfixes, etc ...
Especially the experimental branch is packed with a lot of additional testing 
features. I don't know wether it is wise to ship that ...


> FYI, we sent you 6 merakis today.  I'll get you a tracking number
> shortly.

Great - thanks.

Regards,
Marek