[ovs-discuss] VLAN settings with patch port

[Diego Rivero]

Hi,

Sorry for sending multiple times. What I am concerned is arp request/reply
between br0 and tap0. To send an icmp packet from 10.0.1.2 to 10.0.0.2 I
think the following operations should be done.

1. arp request from 10.0.1.2(tap0) to 10.0.1.1(br0)
2. arp reply from 10.0.1.1(br0) to 10.0.1.2(tap0), so tap0 know the mac
address of br0
3. icmp request sent from 10.0.1.2(tap0) to 10.0.0.2(tap1)
4. br0 forwards the packet to br1 while modifying src mac address and dst
mac address.
5. br1 forwards the packet to tap1 while modifying src and dst mac address.

What I am now facing is that No.2 on the above list is not done
successfully.

If I tag 100 on br0, arp request succeeds. then it starts sending icmp
packet.
If I don't tag 100 on br0, arp request fails.

On your settings, it seems arp request seems to be done successfully.

Is this trivial issue only on Open vSwitch v1.6.1?


On Fri, Feb 22, 2013 at 12:22 PM, Diego Rivero <riverod9 at gmail.com> wrote:

> I am actually using Open vSwitch v1.6.1. I hope this wouldn't be the stem
> of problems.
>
>
>
> On Fri, Feb 22, 2013 at 12:13 PM, Diego Rivero <riverod9 at gmail.com> wrote:
>
>> Hi,
>>
>>
>>
>>> > IP address:
>>> > br0 -> 10.0.1.1
>>> > tap0 -> 10.0.1.2
>>> > br1 -> 10.0.0.1
>>> > tap1 -> 10.0.0.2
>>>
>>> I'm assuming that netmask is 255.0.0.0 (or 255.255.0.0).
>>>
>>>
>> netmask is actually 255.255.255.0. I assume both 10.0.1.0/24 and
>> 10.0.0.0/24 are totally independent. Both networks never know each
>> other. I want to connect those 2 networks temporally.
>>
>> $ sudo ovs-ofctl dump-flows br0NXST_FLOW reply (xid=0x4):
>>>  cookie=0x0, duration=2907.051s, table=0, n_packets=2997,
>>> n_bytes=298734, idle_age=1, priority=0 actions=NORMAL
>>>
>>> $ sudo ovs-ofctl dump-flows br1
>>> NXST_FLOW reply (xid=0x4):
>>>  cookie=0x0, duration=2905.721s, table=0, n_packets=2965,
>>> n_bytes=294622, idle_age=0, priority=0 actions=NORMAL
>>>
>>>
>> I thought packets flow through patch port only with NORMAL flow, but my
>> settings seem not allow to do so (because of netmask). In case this netmask
>> (255.255.255.0), do I need to insert flow like 'priority=100, nw_dst=
>> 10.0.1.0/24, actions=output:#{patch_port_number}' ?
>>
>>
>>> regards
>>> Lucas Brasilino
>>>
>>
>> Thanks,
>>
>
>
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://openvswitch.org/pipermail/ovs-discuss/attachments/20130222/de2ef117/attachment.html>