On 03/11/2014 07:11 AM, Matthew Miller wrote:
On Tue, Mar 11, 2014 at 06:57:13AM -0700, Steven Dake wrote:
After speaking with Steve Baker last night, we discussed
transitioning off cloud-init as a possibility. What we are
considering transitioning to is the os-*-* python tools. The
general idea is these tools would replace cloud-init long term for
the Heat use case. We can't make such changes until Juno wraps up
(November).
Annnd, just throwing this out there: CoreOS uses a partially-compatibile
cloudinit reimplementation, written, of course, in Go:
https://github.com/coreos/coreos-cloudinit
Matt,
Just for further reference, the OpenStack community either understands
there is a problem with agent proliferation, or sees it as a potential
problem. Heat started with 1, and now has to expand to 4 - although 3
of them are generally reusable by TripleO. There is effort underway
upstream to unify the various needs of the agents for services into one
solution that fits everyone's needs. I have never seen efforts like
these succeed before and I've seen ~6 different teams try such a thing,
but maybe this will be a first :)
In the meantime, I think we can expect a bit of pain around
proliferation, at least for the OpenStack use case for guest images. I
know its a pain to have to deal with multiple agents.
Regards
-steve
_______________________________________________
cloud mailing list
[email protected]
https://admin.fedoraproject.org/mailman/listinfo/cloud
Fedora Code of Conduct: http://fedoraproject.org/code-of-conduct
