Open Stack

On Mon, Mar 5, 2018 at 8:07 AM, Matt Riedemann <mriedemos at gmail.com> wrote:
> On 3/5/2018 6:26 AM, Andrey Kurilin wrote:
>> - openstackclient
>>
>> * command `openstack server add floating ip` calls `add_floating_ip`
>> method of novaclient's server object[4]. This method is missed in the latest
>> novaclient (see [2]).
>>    It means that this command is broken now.

>> So here we have 2 global issues:
>> - openstackclient has a broken command (or I missed something?)
>> - there is no easy way to associate a floating ip with a vm using CLI or
>> via python.

> I mentioned the related issue back in January:
>
> http://lists.openstack.org/pipermail/openstack-dev/2018-January/126741.html
>
> Adding a floating IP to an instance is possible using OSC CLI, it's
> essentially something like:
>
> a) get the server id (openstack server show/list)
> b) get the port id using the server id (openstack port list --device-id
> $server_id)
> c) assign the floating IP to the port (openstack floating ip set --port
> $port_id)

<rant>
We keep removing Python API bindings from client libraries that are
still in use for old clouds that are still in much wider use than we
would like.  Why do we not give a rats ass about our users?
Especially when some deployers have multiple clouds lying about,
requiring them to maintain multiple venvs of CLIs is just stupid to be
able to work on their clouds and migrations to the cool new stuff.

OSC is not done because I have about 3 hours a week left to work on
it.  Continued shit like this isn't helping me want to keep going.
Maybe my brain is just snow-fried.   And for the love of all the snow
in Dublin, please, NOBODY USE THE SDK IN A SERVICE.  Keeping service
assumptions out of client-side stuff is the biggest reason OSC NEEDS
to get changed over to the SDK, like, 2 years ago.  Then I'll not give
a rats ass about the legacy python client libs.
</rant>

dt

-- 

Dean Troyer
dtroyer at gmail.com