|
|
|
@ -89,6 +89,25 @@ Deploying a Standalone OpenStack node
|
|
|
|
|
#. Configure basic standalone parameters which include network configuration
|
|
|
|
|
and some deployment options.
|
|
|
|
|
|
|
|
|
|
.. warning::
|
|
|
|
|
The standalone deployment requires one network interface on the deployment
|
|
|
|
|
machine and **that interface will be reconfigured as per the parameters
|
|
|
|
|
you specify below**. The interface you want to use is specified by name
|
|
|
|
|
in the $INTERFACE parameter below and passed as the NeutronPublicInterface
|
|
|
|
|
in the standalone_parameters.yaml. If you only have one interface on your
|
|
|
|
|
machine be advised that it will be reconfigured to have the IP address
|
|
|
|
|
specified in $IP. If that is a remote box you may lose connectivity to it.
|
|
|
|
|
Any other network interfaces are left untouched.
|
|
|
|
|
|
|
|
|
|
For the standalone deployment we use a single NIC on the target machine
|
|
|
|
|
which is reconfigured and set as a member of an ovs bridge, **br-ctlplane**. Two
|
|
|
|
|
examples follow which can be copy/pasted as is - depending on your setup.
|
|
|
|
|
You should only have to change the name of the interface to match whatever
|
|
|
|
|
it is called on your system. Ideally you will have two network interfaces,
|
|
|
|
|
so that one is used for the standalone deployment, whilst the other will be
|
|
|
|
|
left untouched. This can be especially important if you are deploying on a
|
|
|
|
|
remote box (e.g. via ssh).
|
|
|
|
|
|
|
|
|
|
The following configuration can be used for a system with 2 network
|
|
|
|
|
interfaces. This configuration assumes the first interface is used for
|
|
|
|
|
management and we will only configure the second interface. The deployment
|
|
|
|
|