+++ This bug was initially created as a clone of Bug #1890494 +++

We need to clarify the expectations and limitations of egress IP support on clouds in openshift-sdn. (The ovn-kubernetes egress IP documentation is already correct.)

Right now https://docs.openshift.com/container-platform/4.6/networking/openshift_sdn/assigning-egress-ips.html warns:

    Allowing additional IP addresses on the primary network interface might
    require extra configuration when using some cloud or VM solutions.

and later says:

    This approach [manual assignment] is recommended for clusters installed
    in public cloud environments, where there can be limitations on
    associating additional IP addresses with nodes.

But other than that it doesn't say much about bare metal vs private cloud vs public cloud.

What we should be saying is that:

  - The feature is primarily designed for bare-metal and private cloud
    (vSphere/OpenStack/RHEV) environments.

  - Allowing additional IP addresses on the primary network interface
    might require extra configuration when using those private cloud
    environments. (I believe we give more details in the kbase
    somewhere... maybe that should be here?)

  - You can use egress IPs with manual assignment on AWS and Azure if
    you first manually allocate the egress IP to the correct node in the
    AWS/Azure console. (I don't think we want to document how to do that
    in detail... if they can't figure it out, they probably shouldn't be
    using the feature this way. It wasn't _supposed to_ work, it's just
    that it happens to work if you know how to kludge it.)

  - You cannot use egress IPs with automatic assignment on AWS or Azure,
    and you can't use egress IPs in any mode on GCP.