Open Stack

> > > Would we expect users to configure that pipeline?
> 
> > Rarely, it should be more of a side-effect of deployment (e.g.
> > rolling out a CW service would require corresponding updates to
> > the ceilo compute agent config).
> 
> That seems a little brittle. We probably want a set of files, instead
> of a monolithic config, to make it easier to add and remove pieces
> using tools like chef and puppet.


Yep.

In fact, if we stick to the common cfg format, then we get the ability
to pull in config fragments from a conf.d-style directory for free.

(I added this a while back in[1] specifically to support the puppet
use-case).

Cheers,
Eoghan

[1] https://github.com/openstack/oslo-incubator/commit/fd8c2e7c