openstack team mailing list archive

[Vishvananda Ishaya <vishvananda@xxxxxxxxx>]

On Dec 5, 2012, at 7:19 PM, Lei Zhang <zhang.lei.fly@xxxxxxxxx> wrote:

> thank you very much, Vishvananda.
> But I am still confused about the 192.168.0.0/24 and the 10.0.0.0/8 ip. What means by "The addresses will be moved on to the bridge". It means the 192.168.0.0/8 will be disappear?  In my opinion, the bridged NIC (eth1) should be worked under promiscuous mode and its IP should be 0.0.0.0. So the eth1 should not own any IP. 

No moved to the bridge means that the ip will move from eth1 to the bridge

eth1
 -- no ip address

br100
 192.168.0.2
 10.0.0.2

(for example). Nova moves the eth1 ip automatically when it creates the bridge if eth1 has an ip.

Vish

> But if the 192 address doesn't exist,  how the compute-note communicate with each other? Through the eth0? I have no idea. 
> 
> 
> On Thu, Dec 6, 2012 at 3:12 AM, Vishvananda Ishaya <vishvananda@xxxxxxxxx> wrote:
> 
> On Dec 5, 2012, at 1:53 AM, Lei Zhang <zhang.lei.fly@xxxxxxxxx> wrote:
> 
>> Hi all,
>> 
>> I am reading the http://docs.openstack.org/trunk/openstack-compute/admin/content/libvirt-flat-dhcp-networking.html, I got the following deploy architecture. But there are several that I am confused.
>> 
>> How and why 192.168.0.0/24 ip range exist? It is necessary or not? The eth1 on the each physical machine own two ip(10.0.0.0/24 and 192.168.0.0/24)? Is that possible? In the nova-compute, the eth1 should be bridged by br100. the eth1 should not own any IP address, right?
> The addresses will be moved on to the bridge. The point of having an ip address is so that things like rabbit and mysql can communicate over a different set of addresses than the guest network. Usually this would be done on a separate eth device (eth2) or vlan, but I was trying to keep
> 
> 
>> In a better way, should we separate the nova-network/eth0 to the internet public switch for access the internet by all VMs. and the nova-compute/eth0 should be bind to a internal switch for admin access use. Is it right?
> 
> Ideally there are three eth devices / vlans a) public (for 99 adddresses in diagram) b) management (for 192 addresses in diagram) c) guest (for 10 addresses in diagram)
> 
>> 
>> -- 
>> Lei Zhang
>> 
>> Blog: http://jeffrey4l.github.com
>> twitter/weibo: @jeffrey4l
>> 
>> _______________________________________________
>> Mailing list: https://launchpad.net/~openstack
>> Post to     : openstack@xxxxxxxxxxxxxxxxxxx
>> Unsubscribe : https://launchpad.net/~openstack
>> More help   : https://help.launchpad.net/ListHelp
> 
> 
> 
> 
> -- 
> Lei Zhang
> 
> Blog: http://jeffrey4l.github.com
> twitter/weibo: @jeffrey4l
>