IP address configuration support for interfaces in south-ocnos

[pedrosouzajr]

This issue discusses how to support the IP address configuration when we have the OcNOS available.
The proposal to support this configuration considers adding a module called goldstone-ocnos making an augment of the goldstone-interfaces module as displayed below:

• goldstone-ocnos.yang yang-file

module: goldstone-ocnos
  augment /gs-if:interfaces/gs-if:interface:

    +--rw ipv4
       +--rw addresses
       |  +--rw address* [ip]
       |     +--rw ip        -> ../config/ip
       |     +--rw config
       |     |  +--rw ip?              inet:ipv4-address
       |     |  +--rw prefix-length?   uint8
       |     +--ro state
       |        +--ro ip?              inet:ipv4-address
       |        +--ro prefix-length?   uint8
       |        +--ro origin?          ip-address-origin
       +--rw config
       +--ro state

Please, notice that neighbors container inside goldstone-ipv4 is not considered in this first step.

[ishidawataru]

Do we need to consider if we can assign an IPv4 address to an interface when the interface belongs to one or more VLANs?

[ipi-claytonpascoal]

Do we need to consider if we can assign an IPv4 address to an interface when the interface belongs to one or more VLANs?

Thank you for pointing that out.
Yes, on OcNOS an interface is either a switch port or a router port so it is not allowed to assign an IP address to an interface belonging to a VLAN
Kindly asking @pedrosouzajr to consider adding a constraint to cover this case.

[pedrosouzajr]

Thank you all for pointing that out.
So, to attend to this constraint which was suggested the goldstone-ocnos.yang was changed to add a 'must' statement for this purpose.
Please, consider this new version: goldstone-ocnos.yang

[ipi-claytonpascoal]

@ishidawataru @anazawa-kazuya
Please notice that we are not willing to support /gs-if:interfaces/interface/gs-ocnos:ipv4/neighbors in this first step.
Can we move on and create a pull request including the goldstone-ocnos module as proposed in this discussion?

[KazuyaAnazawa]

@pedrosouzajr @ipi-claytonpascoal I think we can proceed with the model you proposed (though we may need to fix or add some minor parts during actual implementation of south-ocnos). Then, we can finalize the model while implementing south-ocnos as comments here.
Also, let's create and submit a PR of goldstone-ocnos.yang with actual implementation of IP handler part of south-ocnos after we completed it.