I concur. The only thing with RFE's and patch approvals is I think we should remember that we want it to be easy. So processes like RFEs may not be helpful to a "oh, this tiny little thing makes a lot of sense" sort of things, since it quickly becomes a situation where you spend more time on the RFE than the patch itself. On Mon, Feb 8, 2021 at 7:43 AM Ruby Loo <opensrloo@gmail.com> wrote:
Hi Dmitry,
Thanks for bringing this up! We discussed this in our weekly ironic meeting [1]. The consensus there seems to be to keep the ideas in IPA (with priority=0). The additional code will be 'negligible' in size so ramdisk won't be bloated due to this. Also, it keeps things simple. Having a separate package means more maintenance overhead and confusion for our users.
Would be good to hear from others, if they don't think this is a good idea. Otherwise, I'm looking forward to Dmitry's RFEs on this :)
--ruby
[1] http://eavesdrop.openstack.org/irclogs/%23openstack-ironic/%23openstack-iron...
On Mon, Feb 8, 2021 at 8:02 AM Dmitry Tantsur <dtantsur@redhat.com> wrote:
Hi all,
We have finally implemented in-band deploy steps (w00t!), and people started coming up with ideas. I have two currently: 1) configure arbitrary kernel command line arguments via grub 2) write NetworkManager configuration (for those not using cloud-init)
I'm not sure how I feel about putting these in IPA proper, seems like we may go down a rabbit hole here. But what about a new project (ironic-python-agent-extras?) with a hardware manager providing a collection of potentially useful deploy steps?
Or should we nonetheless just put them in IPA with priority=0?
Opinions welcome.
Dmitry
-- Red Hat GmbH, https://de.redhat.com/ , Registered seat: Grasbrunn, Commercial register: Amtsgericht Muenchen, HRB 153243, Managing Directors: Charles Cachera, Brian Klemm, Laurie Krebs, Michael O'Neill