Floating IP , External network relationship

Asked by Akilesh

I am a network guy and the concepts of floatingip and external network looks puzzling to me.

The floating ip is allocated from the allocation pool of the subnet created in the external network. What happens if the external network has two subnets.

A similar conflict of logic arises when lauching instances. You can specify which network the instance is launched in but not which subnet if the network has more than one.

Why does openstack neutron allow this to happen.
Why have a separate entity called network?
and If we do have why allow creation of multiple subnets in them?

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
yong sheng gong (gongysh) said :
#1

IMHO, network is the abstract of the whole network space in physical realitity. In general one tenant needs only one neutron network. With one network, tenant can create subnets just like traditional ways as normal.

regarding floatingip's subnets, it depends on ip allocation algorithm in neutron. currently, we will choose a IP randomly from all subnets in the external network.

for launching instance, you can specify the IP address, which equals to select subnet.

hope this helps.

Revision history for this message
Akilesh (akilesh1597) said :
#2

I recognize the significance but there are still some issues.

for launching instance, specifying the IP address, can not be equal to select subnet because you can not tell openstack that you want a dhcp allocated address from a subnet. further it would be much easier to select a subnet.

Also does the logic guarantee that the random subnet chosen to create floatingip is the one connected to the tenant subnet by a router.

Revision history for this message
yong sheng gong (gongysh) said :
#3

many guys are confused by dhcp role in neutron. In neutron, the dhcp server (dnsmasq) is a guided one, I.E. neutron will wirte the IP and mac pair into dnsmasq's configure file, and then dnsmasq can just allocate the IP according to the requester's mac. So the instance's IP is totally controlled by neutron, instead of dnsmasq.

I think floatingip must be along with the router's gateway port's IP at the same subnet. i.e. the system should not select subnet ramdonly for floatingip. I have to check wether it is a bug.

Revision history for this message
Akilesh (akilesh1597) said :
#4

I believe you do not understand the problem. I understand that dhcp address is controlled by neutron. I was talking about giving users the ability to choose subnets rather than use a random subnet from the user's network.

Can you help with this problem?

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

To post a message you must log in.