Ask Your Question

Revision history [back]

click to hide/show revision 1
initial version

Network subnet help with tenant and management

In openstack we have a tenant interface on the compute node but why do we need to also present this network to the controller?

As I understand the tenant network(s) give the instances access to the networks external to openstack. Since no instances run on the controller I am uncertain why this is required there.

Additionally, we have a management network which I've read is required for the ml2 plugin to communicate with the real network hardware when using the integration. But why would this network be required on compute nodes?

As I understand this network is used by the controller to make changes to the underlying network for the compute nodes instances. But the compute node itself doesnt make these changes.

So it seems apparent to me that the tenant network is required on the compute node only and the management network is required on the controller only. But all documentation and guides state both networks are required on both nodes and doesn't seem correct.

Please can someone shine light on this?