neutron

VM is loosing it's IP address with the addition of flows in the bridge

Asked by NAVEEN REDDY

Hi all,

We have enabled vxlan-tunneling on both the controller and compute hosts, we were able to see br-tun in a down state by default.So, we have bring br-tun in up state.

By default we are seeing vxlan-0 is attached in br-tun bridge and all the VM interfaces and eth1 are attached in br-int bridge.

With this setup, we are unable to ping from VM[controller] to VM[compute].

Then we tried in the following way:
1. Removed vxlan-0 port from br-tun bridge
2. Removed eth1 port from br-int bridge
3. Added vxlan-0 port to br-int bridge, because all VM interfaces are attached in the br-int bridge
   ovs-vsctl add-port br-int vxlan-0 -- set Interface vxlan-0 type=vxlan options:remote_ip=10.1.1.1

Above 3 steps we have tried in both the controller and compute hosts.
Remote_ip which we mentioned in step 3 is eth1 ip address.

Now we are able to ping between VM[controller] and VM[Compute].
we captured packets on eth1 on both controller and compute, we observed all the packets are going on eth1.

Can you please tell us is this correct way of doing or something else is missing from our side or it is already known issue[bug]?

We have observed one more intersting problem i.e

we have created two different networks, one is 20.0.0.0 and 30.0.0.0
On Compute we have VM's with IP addresses are 20.0.0.2 and 30.0.0.2
On Controller we have VM's with IP addresses are 20.0.0.3 and 30.0.0.3

we are able to ping between 20.0.0.2 and 30.0.0.2, but not able to ping between 20.0.0.2 and 30.0.0.3
[Above scenario is happening on Compute hosts]

we are not able to ping between 20.0.0.3 and 30.0.0.3, but able to ping between 20.0.0.3 and 30.0.0.2
[Above scenario is happening on Controller hosts]

Can you please confirm us, is this already a known issue[bug] or not ?

Another problem which we have observed is whenever we add flows in br-int bridge , VM is loosing it's IP address.

Example of flows:
        ovs-ofctl add-flow br-int in_port=1,actions=output:3
        ovs-ofctl add-flow br-int in_port=3,actions=output:1

1 = VM's port
3 = vxlan-0 port

Thanks in advance,
Naveen Reddy

Question information

Language:
English Edit question
Status:
Answered
For:
neutron Edit question
Assignee:
No assignee Edit question
Last query:
Last reply:
Revision history for this message
dan wendlandt (danwent) said : 		#1

Hi Naveen,

I need to get a repro environment setup to test this. Also, what version
of OVS are you using? I should be able to look at this over the weekend.

dan

On Thu, Apr 26, 2012 at 3:20 AM, NAVEEN REDDY <
<email address hidden>> wrote:

> New question #194777 on quantum:
> https://answers.launchpad.net/quantum/+question/194777
>
> Hi all,
>
> We have enabled vxlan-tunneling on both the controller and compute hosts,
> we were able to see br-tun in a down state by default.So, we have bring
> br-tun in up state.
>
> By default we are seeing vxlan-0 is attached in br-tun bridge and all the
> VM interfaces and eth1 are attached in br-int bridge.
>
> With this setup, we are unable to ping from VM[controller] to VM[compute].
>
> Then we tried in the following way:
> 1. Removed vxlan-0 port from br-tun bridge
> 2. Removed eth1 port from br-int bridge
> 3. Added vxlan-0 port to br-int bridge, because all VM interfaces are
> attached in the br-int bridge
> ovs-vsctl add-port br-int vxlan-0 -- set Interface vxlan-0 type=vxlan
> options:remote_ip=10.1.1.1
>
> Above 3 steps we have tried in both the controller and compute hosts.
> Remote_ip which we mentioned in step 3 is eth1 ip address.
>
> Now we are able to ping between VM[controller] and VM[Compute].
> we captured packets on eth1 on both controller and compute, we observed
> all the packets are going on eth1.
>
> Can you please tell us is this correct way of doing or something else is
> missing from our side or it is already known issue[bug]?
>
> We have observed one more intersting problem i.e
>
> we have created two different networks, one is 20.0.0.0 and 30.0.0.0
> On Compute we have VM's with IP addresses are 20.0.0.2 and 30.0.0.2
> On Controller we have VM's with IP addresses are 20.0.0.3 and 30.0.0.3
>
> we are able to ping between 20.0.0.2 and 30.0.0.2, but not able to ping
> between 20.0.0.2 and 30.0.0.3
> [Above scenario is happening on Compute hosts]
>
> we are not able to ping between 20.0.0.3 and 30.0.0.3, but able to ping
> between 20.0.0.3 and 30.0.0.2
> [Above scenario is happening on Controller hosts]
>
> Can you please confirm us, is this already a known issue[bug] or not ?
>
> Another problem which we have observed is whenever we add flows in br-int
> bridge , VM is loosing it's IP address.
>
> Example of flows:
> ovs-ofctl add-flow br-int in_port=1,actions=output:3
> ovs-ofctl add-flow br-int in_port=3,actions=output:1
>
> 1 = VM's port
> 3 = vxlan-0 port
>
> Thanks in advance,
> Naveen Reddy
>
>
>
> --
> You received this question notification because you are an answer
> contact for quantum.
>

--
~~~~~~~~~~~~~~~~~~~~~~~~~~~
Dan Wendlandt
Nicira, Inc: www.nicira.com
twitter: danwendlandt
~~~~~~~~~~~~~~~~~~~~~~~~~~~

Can you help with this problem?

Provide an answer of your own, or ask NAVEEN REDDY for more information if necessary.

To post a message you must log in.