Open Stack

> I'm going to have to agree with Jay - and I believe with something Clint
> said earlier.
> 
> Use metadata/config-drive to bootstrap you. Then use your actual tools
> (puppet/chef/ansible/salt/juju/cfengine) to interact with the node once
> it's enrolled in whatever you use to orchestrate cloud things. The best
> part about doing that is that your orchestration layer can talk to the
> OpenStack APIs AND to your nodes. Most of them also support key-value
> pairs, as well as updating them.

Yep, agree as well. I hate that metadata is changeable, but I'm not sure
there is much we can do about it now because I think at least one
"feature" is implemented this way. However, definitely +1 to not adding
any more such pain.

We've seen plenty of issues around scaling (read: puppet DoSing) the
metadata stuff. Caching it forever as a static file and/or using
configdrive makes those operations so much more efficient, especially in
the face of dumb things like facter constantly banging the crap out of it.

Many people ask us to reimplement things like auth, nagios, and config
management in nova. That doesn't mean it's something we should do.

--Dan