[RFE] Create a full load balancing configuration with one API call

[Brandon Logan]

There have been many requests to allow the API to expose the ability to create a load balancer, listener, pool, member, and health monitor in one single API request. Some reasons for this:

1) Drivers will know immediately up front what resources will be required to satisfy the end configuration. They can make more efficient/optimized decisions based on this.

2) It's not a good UX to have a user make at minimum 4 API requests before they can actually have a load balancer.

3) Reducing the number of API requests will improve performance of the API server when at large scale and other services consuming the same API.

[Doug Wiegley (dougwig)]

Approved.

[Armando Migliaccio (armando-migliaccio)]

https://review.openstack.org/#/q/owner:%22Brandon+Logan+%253Cbrandon.logan%2540rackspace.com%253E%22+status:open

[Armando Migliaccio (armando-migliaccio)]

status update: http://eavesdrop.openstack.org/meetings/neutron_drivers/2016/neutron_drivers.2016-01-28-22.20.log.txt

[Brandon Logan (brandon-logan)]

update on this. I've agreed its probably best to let the L7 work go in before this because it'll be more work for them to also have to add that work into this. This now looks doubtful to make it in Mitaka because of this.

[Brandon Logan (brandon-logan)]

I wasn't clear in the last comment. Basically, I'm also going to add the L7 functionality to this as well so that L7 can go in before this.

[Armando Migliaccio (armando-migliaccio)]

Sorry, your comments are still cryptic to. Are you saying that it's unlikely to deliver the 'get-me-a-loadbalancer' functionality in time for Mitaka?

[Brandon Logan (brandon-logan)]

yes, it probably won't be able to make it in for MItaka since I'm being realistic that the L7 patch is going to take the majority of the time left to get in. Since I've agreed I'll add the L7 functionality into the get-me-a-lb feature, it is now dependent on L7 getting in. Otherwise, it'd be ready for review. Having said that, there's still a chance it does, just a low probability.

[OpenStack Infra (hudson-openstack)]

Reviewed: https://review.openstack.org/257201
Committed: https://git.openstack.org/cgit/openstack/neutron-lbaas/commit/?id=eafbfd23b246b2b9828cdeb00f2acf7cb0ac0aca
Submitter: Jenkins
Branch: master

commit eafbfd23b246b2b9828cdeb00f2acf7cb0ac0aca
Author: Trevor Vardeman <email address hidden>
Date: Tue Mar 1 14:52:33 2016 -0600

    Get Me A LB

    Allows a request to be passed to the lbaas API that contains the full graph
    structure of a load balancer. This means that an entire load balancer graph
    can be created in one POST to the API, or a partial graph can be.

    This creates a new driver method to the driver interface that when implemented
    will be called when a full or partial graph is attempted to be created. If
    the driver does not implement this method, then the request is not allowed.

    Co-Author: Trevor Vardeman <email address hidden>

    Closes-Bug: #1463202
    Depends-On: Id3a5ddb8efded8c6ad72a7118424ec01c777318d
    Change-Id: Ic16ca4cb3cb407d9e91eea74315a39bf86c654f3

[Doug Hellmann (doug-hellmann)]

This issue was fixed in the openstack/neutron-lbaas 9.0.0.0b2 development milestone.