[ovs-discuss] Bridge pinging rules

Amer amer7777 at hotmail.com
Sat Mar 14 17:09:33 UTC 2015


nano /etc/network/interfaces
auto br0
Inface br0 inet static
address 192.168.56.4
netmask 255.255.255.0
.....

then in the host
gateway 192.168.56.4

Thank you

‏‫‏Ben Pfaff ‏<blp at nicira.com>:‬

> No.
> 
> The local port is the port with the same name as the bridge itself.
> 
>> On Sat, Mar 14, 2015 at 07:11:42PM +0300, Amer wrote:
>> Do you want me to give one of the Ethernet interfaces an IP address= the bridge IP, without assigning this interface to the bridge.
>> 
>> Best regards,
>> Amer
>> 
>> م، كتب ‏Ben Pfaff ‏<blp at nicira.com>:‬
>> Well, like you said, without flows, ping works.  So pass through the
>>> ICMP traffic to the gateway to the IP stack (presumably the bridge's
>>> local port) and let it do the ICMP responses.
>>> 
>>>> On Sat, Mar 14, 2015 at 06:40:10PM +0300, Amer wrote:
>>>> Thank you Ben
>>>> The problem is the host need to have a gateway, however, if he can't ping the gateway, nothing will be forwarded.
>>>> However, how can I pass the ICMP packets to the IP stack?
>>>> 
>>>> Best regards,
>>>> Amer
>>>> 
>>>> ‏:‬
>>>> 
>>>>>> On Sat, Mar 14, 2015 at 02:29:07PM +0300, Amer wrote:
>>>>>> I have 4 VMs, 2 bridges and each one has a host.
>>>>>> My question is:
>>>>>> I need to configure a default gateway for the hosts, however when I added flow rules to the bridges, suddenly, I can't ping them from the hosts. I know this is due to the rules.
>>>>>> When the host can't ping the bridge, how I configure it to be the gateway.
>>>>>> I need to add a flow-rule to make the bridge respond to ping requests? What is the rule to do so?
>>>>> 
>>>>> I don't think you can do that through flows.  It is probably better to
>>>>> configure the flows so that anything that needs an ICMP response is
>>>>> passed through to the IP stack.
> 
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://openvswitch.org/pipermail/ovs-discuss/attachments/20150314/b870d428/attachment-0002.html>


More information about the discuss mailing list