Re: [atomic-devel] Kubernetes in containers

[Colin Walters <walters verbum org>]

On Wed, Sep 16, 2015, at 07:23 PM, Clayton Coleman wrote:

> We don't want to encourage local host paths in the future.  For now it's fine

Are you planning to use Docker volumes then?  Or how else to enable persistence
in the single-host scenario?

(reads farther down)

> The bulk will move into etcd and be defaulted, the remainder
> (connection info, admin secrets) is probably going to have to come
> from the host.  There will probably have to be a "slurp in config"
> option on master startup forever.

Interesting.  Hmm.  And etcd will run as a separate container then I'm
assuming, rather than the all-in-one? If we're storing private keys in etcd,
then that needs to be secured, which means we need keys...

Maybe we say that for production use, we expect to use an external
CA, and for non-production use (e.g. local Vagrant dev), the defaults
are insecure, and we expect your VM to be behind NAT or a firewall.

> It's going to be a while before Kube supports pure additive function.

Yeah, I understand that.  I think broadly speaking too what OpenShift
has been doing with shipping earlier versions of some important
features is good - it gets them more real-world use before they go
upstream.

> similar to https://github.com/openshift/origin/blob/master/examples/ha/openshift-ha.yaml).

Wait...is that running OpenShift inside an existing Kubernetes cluster,
or is it a self-referential OpenShift?