[Openstack] Problem with floating IP

Krzysztof G krzysiekg18 at gmail.com
Wed Sep 4 13:51:29 UTC 2013


Hi everybody,

I'm trying to deploy Openstack with Quantum networking on one node. I did
all-in-one installation, then disabled nova-networking and installed
Quantum networking. I configured internal networking with network address:
192.168.32.0/24. My host is connected to 10.133.31.128/27 company network
in which I have few empty addresses which can be assigned to my VM's. I
used my eth0 interface on host (which is in company network) as an external
bridge and my eth1 as an internal bridge for VM's.
I was following configuration written in this topic
http://openstack.redhat.com/forum/discussion/comment/611

My topology is:

http://s8.postimg.org/iaastbzx1/topology.jpg

When I do 2 VM I can successfully assigned them with Internal addresses an
also floating IP's. I can ping between them using internal address, but I
can't ping my br-int address and cannot ping VM's from my host,  floating
addresses are useless.

I cannot understand one thing. When I'm on server SV1 and want to ping VM1.
Theoretically they are on the same subnet (10.133.31.128/27, beacuse I use
free adresses from this subnet). But the real path is by the **br-ex and
then OpenStack router**. How my SV1 and also switch SW know the correct
path to VM1? Is floating IP telling the switch that the path to VM1 is on
the same interface as it can reach br-ex (eth0) interface?

Here is screenshot of my instances:

http://s11.postimg.org/5btsqsfzm/instances.jpg

and my network topology:

http://s21.postimg.org/dkfhbgepj/network.jpg

Thank you for your help.
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://lists.openstack.org/pipermail/openstack/attachments/20130904/3df2b671/attachment.html>


More information about the Openstack mailing list