Re: Strange ARP problem

Philippe Latu <[email protected]>
Newsgroups gmane.linux.drivers.vlan
Organization Linux France
Message-ID <[email protected]> 
Hello,

My suggestion may be completely off topic, but as you said, any suggestion is 
welcome ...

Do you know that a Cisco interface has proxy-arp enabled by default ?

You should set 'no proxy-arp' on interface GigabitEthernet4/13.

HTH,

Le Jeudi 26 Mai 2005 15:55, Normando Marcolongo a écrit :
> Justin Cormack wrote:
> > I think you have to do it while the interface is down.
>
> Yeah I did but the problem persists. The fact is that Cisco too has the
> same MAC across multiple VLAN:
>
> # arp -an
> ? (172.19.1.1) at 00:07:B3:1A:AC:0A [ether] on eth0
> ? (172.22.1.1) at 00:07:B3:1A:AC:0A [ether] on eth0.201
>
> There you can see the linux machine catching the client broadcasts on
> the eth0.201:
>
> 09:48:39.392575 IP 172.22.1.5.137 > 172.22.1.255.137: NBT UDP
> PACKET(137): REGISTRATION; REQUEST; BROADCAST
> 09:48:40.128021 IP 172.22.1.5.137 > 172.22.1.255.137: NBT UDP
> PACKET(137): REGISTRATION; REQUEST; BROADCAST
> 09:48:40.688972 00:03:fe:ac:09:58 > 01:00:0c:cc:cc:cd sap aa ui/C len=39
> 09:48:40.877931 IP 172.22.1.5.137 > 172.22.1.255.137: NBT UDP
> PACKET(137): REGISTRATION; REQUEST; BROADCAST
> 09:48:41.627924 IP 172.22.1.5.137 > 172.22.1.255.137: NBT UDP
> PACKET(137): REGISTRATION; REQUEST; BROADCAST
> 09:48:42.378053 IP 172.22.1.5.137 > 172.22.1.255.137: NBT UDP
> PACKET(137): REGISTRATION; REQUEST; BROADCAST
> 09:48:42.688993 00:03:fe:ac:09:58 > 01:00:0c:cc:cc:cd sap aa ui/C len=39
> 09:48:43.127830 IP 172.22.1.5.137 > 172.22.1.255.137: NBT UDP
> PACKET(137): REGISTRATION; REQUEST; BROADCAST
> 09:48:43.877744 IP 172.22.1.5.137 > 172.22.1.255.137: NBT UDP
> PACKET(137): REGISTRATION; REQUEST; BROADCAST
>
> and there you can see the client (.5) asking the MAC address of the
> server (.2) that replies. The reply is _not_ arriving on the client cable.
>
> 09:49:10.034136 arp who-has 172.22.1.2 tell 172.22.1.5
> 09:49:10.034159 arp reply 172.22.1.2 is-at 06:0b:db:a9:0f:d1
>
> Note the mac address that is the correct one:
>
> # ifconfig
> eth0      Link encap:Ethernet  HWaddr 00:0B:DB:A9:0F:D1
>           inet addr:172.19.1.11  Bcast:172.19.1.255  Mask:255.255.255.0
>           inet6 addr: fe80::20b:dbff:fea9:fd1/64 Scope:Link
>           UP BROADCAST RUNNING PROMISC MULTICAST  MTU:1500  Metric:1
>           RX packets:299499 errors:0 dropped:0 overruns:0 frame:0
>           TX packets:161893 errors:0 dropped:0 overruns:0 carrier:0
>           collisions:0 txqueuelen:1000
>           RX bytes:22443458 (21.4 MiB)  TX bytes:220560315 (210.3 MiB)
>           Base address:0xece0 Memory:fe8e0000-fe900000
>
> eth0.201  Link encap:Ethernet  HWaddr 06:0B:DB:A9:0F:D1
>           inet addr:172.22.1.2  Bcast:172.22.1.255  Mask:255.255.255.0
>           inet6 addr: fe80::40b:dbff:fea9:fd1/64 Scope:Link
>           UP BROADCAST RUNNING MULTICAST  MTU:1500  Metric:1
>           RX packets:78326 errors:0 dropped:0 overruns:0 frame:0
>           TX packets:25 errors:0 dropped:0 overruns:0 carrier:0
>           collisions:0 txqueuelen:0
>           RX bytes:3923165 (3.7 MiB)  TX bytes:1610 (1.5 KiB)
>
> I'm going to give up... Please if anyone has a suggestion is very welcome.
>
> Thanks,
> Normando
>
> > Justin
> >
> > On 24 May 2005, at 12:30, Normando Marcolongo wrote:
> >> Justin Cormack wrote:
> >>> I think that lots of equipment doesnt like you mixing tagged and
> >>> untagged traffic on the same network.
> >>>
> >>> If you have to it might help setting the Mac addresses on the two
> >>> interfaces to be different.
> >>
> >> So how can I change the mac on the eth0.201 interface?
> >>
> >> # ifconfig eth0.201 hw ether 00:0B:DB:A9:1F:D1
> >> SIOCSIFHWADDR: Device or resource busy
> >>
> >> Thanks!!!
> >> Normando
> >>
> >>> On 24 May 2005, at 11:32, Normando Marcolongo wrote:
> >>>> Hi!
> >>>>
> >>>> Let me explain the simple architecture here. A linux machine  with
> >>>> kernel 2.6.11.7 (Debian Sarge) and e1000 driver. A Cisco  Catalyst
> >>>> 6509 configured with a trunk port this way:
> >>>>
> >>>> interface GigabitEthernet4/13
> >>>>  switchport
> >>>>  switchport trunk native vlan 100
> >>>>  switchport trunk allowed vlan 100,201
> >>>>  switchport mode trunk
> >>>>
> >>>> The linux machine has:
> >>>>
> >>>> # cat /proc/net/vlan/config
> >>>> VLAN Dev name    | VLAN ID
> >>>> Name-Type: VLAN_NAME_TYPE_RAW_PLUS_VID_NO_PAD
> >>>> eth0.201       | 201  | eth0
> >>>>
> >>>> # ifconfig
> >>>> eth0      Link encap:Ethernet  HWaddr 00:0B:DB:A9:0F:D1
> >>>>           inet addr:172.19.1.11  Bcast:172.19.1.255  Mask:
> >>>> 255.255.255.0
> >>>>           inet6 addr: fe80::20b:dbff:fea9:fd1/64 Scope:Link
> >>>>           UP BROADCAST RUNNING MULTICAST  MTU:1500  Metric:1
> >>>>           RX packets:49984 errors:0 dropped:0 overruns:0 frame:0
> >>>>           TX packets:80330 errors:0 dropped:0 overruns:0 carrier:0
> >>>>           collisions:0 txqueuelen:1000
> >>>>           RX bytes:4335305 (4.1 MiB)  TX bytes:110105417 (105.0 MiB)
> >>>>           Base address:0xece0 Memory:fe8e0000-fe900000
> >>>>
> >>>> eth0.201  Link encap:Ethernet  HWaddr 00:0B:DB:A9:0F:D1
> >>>>           inet addr:172.22.1.2  Bcast:172.22.1.255  Mask:
> >>>> 255.255.255.0
> >>>>           inet6 addr: fe80::20b:dbff:fea9:fd1/64 Scope:Link
> >>>>           UP BROADCAST RUNNING MULTICAST  MTU:1500  Metric:1
> >>>>           RX packets:255 errors:0 dropped:0 overruns:0 frame:0
> >>>>           TX packets:361 errors:0 dropped:0 overruns:0 carrier:0
> >>>>           collisions:0 txqueuelen:0
> >>>>           RX bytes:22471 (21.9 KiB)  TX bytes:21547 (21.0 KiB)
> >>>>
> >>>> If I put a test client on the VLAN100 the linux machine can ping
> >>>> the 6509's VLAN100 interface and the test client.
> >>>> Now, I can ping from the linux machine the VLAN201 interface of
> >>>> the  6509 and (obviuosly) vice versa.
> >>>> If I put a test client on the VLAN201 the test client _cannot_
> >>>> resolve the MAC address of the linux machine while the linux
> >>>> machine has this resolved. If I do a tcpdump on the linux machine
> >>>> I  see:
> >>>> on the eth0.201 the ARP request from the client _and_ the reply
> >>>> from the linux machine to the client (this will never reach the
> >>>> client for an unknown reason)
> >>>> on the eth0 the same (I think I've understood that this is normal)
> >>>> putting static ARPs on the server/client solves the problem.
> >>>> Can you give me some directions? Need some other information?
> >>>>
> >>>> Thanks,
> >>>> Normando
> >>>> -- Normando Marcolongo | Micso s.r.l. via Tiburtina, 318 |
> >>>> I-65128  Pescara, Italy tel/fax (+39)08554105 | mob. (+39)3386296362
> >>>> <normando.vcf>
> >>>> _______________________________________________
> >>>> Vlan mailing list
> >>>> [email protected]
> >>>> http://www.lanforge.com/mailman/listinfo/vlan
> >>>
> >>> _______________________________________________
> >>> Vlan mailing list
> >>> [email protected]
> >>> http://www.lanforge.com/mailman/listinfo/vlan
> >>
> >> --
> >>  Normando Marcolongo | Micso s.r.l.
> >>   via Tiburtina, 318 | I-65128 Pescara, Italy
> >> tel/fax (+39)08554105 | mob. (+39)3386296362
> >>
> >>
> >> <normando.vcf>
> >> _______________________________________________
> >> Vlan mailing list
> >> [email protected]
> >> http://www.lanforge.com/mailman/listinfo/vlan
> >
> > _______________________________________________
> > Vlan mailing list
> > [email protected]
> > http://www.lanforge.com/mailman/listinfo/vlan

-- 
- Philippe Latu
< G N U / Linux >
philippe.latu(at)linux-france.org
Projet inetdoc.Linux 
http://www.linux-france.org/prj/inetdoc
</>
< I U T 'A'  Paul Sabatier >
philippe.latu(at)iut-tlse3.fr - 05.62.25.80.28
Enseignant/Chargé de mission Systèmes & Réseau
</>
lmpx.com only provides a reader for public news (NNTP) servers. It is not affiliated with the servers or forums shown here and is not responsible for the content of articles, which is written by their respective authors.