OpenStack Compute (nova)

network interfaces confusion

Asked by amy lee

Hi,

I am new to nova and am struggling to get the proper understanding about the networking part in it. I couldn't find information regarding it in the documentation.

so my original interface is eth0.
After installing nova I have a bridge br100 which gets my ip from eth0 - so I guess this is the interface to communicate with the rest of my non-cloud network. at least it works that way right now.

I also get a virbr0 interface which has the ip 192.168.122.1 - what is this used for? nowhere in the configuration I used a 192-network, so I am confused about that. Also I have vnet0.

Which interface do I need to configure on my host to communicate with the running instances - the virbr0 or the vnet0?

I tried both of them without success... but right now I'd be happy to at least get the proper understanding how it theoretically should work.

Thanks in advance,
Amy

Question information

Language:
English Edit question
Status:
Answered
For:
OpenStack Compute (nova) Edit question
Assignee:
No assignee Edit question
Last query:
Last reply:
Revision history for this message
Soren Hansen (soren) said : 		#1

virbr0 is created automatically by libvirt. It's not related to Nova.

What do you mean by "I tried both of them without success... "? What did you try/do exactly?

Revision history for this message
amy lee (amylee) said : 		#2

according to euca-describe-instances and the networks table in the database my instance got the IP 10.1.42.2/25 (I also already tried mounting the image and putting that IP directly into the config file)

so I gave vnet0 the IP 10.1.42.1/25 and tried to reach the instance via ping and ssh.
I enabled both first with:
euca-authorize -P icmp -t -1:-1 default
euca-authorize -P tcp -p 22 default

and then I changed it and gave virbr0 the IP 10.1.42.1/25 instead.

Or would I have to change my br100 IP instead?

which interface would I use to communicate with my instances?

Revision history for this message
Soren Hansen (soren) said : 		#3

You don't have to (and shouldn't!) do any of that. The instance is configured (either by DHCP or injection, depending on your network model) with the IP. None of the interfaces you see on the host will have this IP.

Revision history for this message
amy lee (amylee) said : 		#4

yes, the instance is configured... but to communicate with it, my cloud controller needs to be in the same network, right?! Doesn't my hosting system need some interface in the same network as the instance?!

Revision history for this message
Soren Hansen (soren) said : 		#5

2011/4/5 amy lee <email address hidden>:
> yes, the instance is configured... but to communicate with it, my cloud
> controller needs to be in the same network, right?! Doesn't my hosting
> system need some interface in the same network as the instance?!

The network controller should at least have an IP in the subnet, yeah.

Doesn't it? Which network manager are you using?

Revision history for this message
amy lee (amylee) said : 		#6

network manager? well I use ifconfig and route...

still my question would be: which interface of the host system should be in the subnet?

Revision history for this message
Hyunsun Moon (hyunsun-moon) said : 		#7

Hi, check network_manager in nova.conf file. It's on $NOVA_DIR/bin/ or /etc if you installed with deploy tool.
I guess it's VlanManager if you haven't changed it.

In that case, nova-network fetches IP address for VM from dnsmasq at instance creation.
You shouldn't assign IP address for network interface of VM on your host OS.

Just run instance and check the IP address displayed as a result of euca-describe-instance.
You can ping or ssh to the VM with that IP address.
No additional settings are required.

If you're Korean, visit www.openstack.or.kr :)

Can you help with this problem?

Provide an answer of your own, or ask amy lee for more information if necessary.

To post a message you must log in.