Open Stack

gordon chung wrote:
>
> On 11/10/16 05:38 PM, Joshua Harlow wrote:
>> Yes yes, normalization would be nice to, though a little beyond what I
>> am (or was) thinking of currently. Going back to how
>> event_definitions.yaml is the best 'source' we have currently, is it
>> possible to rip out (for now) event_definitions.yaml  and I assume the
>> python code-gen part from ceilometer into a new library (and/or repo);
>> at least this would be a small step forward (I think).
>>
>
> i don't have a problem with it but i'll be honest, the definition file
> and yaml->python object code is tailored to an extent to "normalise
> notification into Ceilometer data models" use case.

Right, I'm assuming this is primarily what's in:

https://github.com/openstack/ceilometer/blob/master/ceilometer/event/

That and the event_definitions.yaml seems like they could just be in a 
new location (not sure what to call that location, someone else can pick 
a cool name, ha).

>
>> Perhaps this may have already been planned to?
>
> we did plan on making Ceilometer notification agent more generic where
> it's essentially just a notification listener and you can enable
> different plugins to convert to meters or events where as right now they
> are both pretty much baked in and always on. maybe this generalisation
> can be tied into what you are tryin to do.
>
>

Right this is one option, though it still ties people into ceilometer 
for events, which feels odd, vs just being a generic tool/library/tiny 
entrypoint that could be provided without needing ceilometer (that 
ceilometer could itself also use).

-Josh