[openstack-dev] [magnum] The way magnum-conductor communicates with k8s master
Steven Dake (stdake)
stdake at cisco.com
Tue Jul 14 03:40:51 UTC 2015
From: "OTSUKA, Motohiro" <yuanying at oeilvert.org<mailto:yuanying at oeilvert.org>>
Reply-To: "OpenStack Development Mailing List (not for usage questions)" <openstack-dev at lists.openstack.org<mailto:openstack-dev at lists.openstack.org>>
Date: Monday, July 13, 2015 at 8:11 PM
To: "OpenStack Development Mailing List (not for usage questions)" <openstack-dev at lists.openstack.org<mailto:openstack-dev at lists.openstack.org>>
Subject: Re: [openstack-dev] [magnum] The way magnum-conductor communicates with k8s master
Currently magnum-conductor can communicates with k8s master which has a floating ip in all-in-one deployment. But if magnum-conductor is not deployed on the neutron network node which has the br-ex, how can magnum-conductor communicate with k8s master. The magnum-conductor node then has no access to the k8s master.
Currently this is a limitation of our architecture.
The floating IP is a public routed network, so it should be able to communicate in a properly setup cloud.
Magnum-conductor only communicate with k8s master, why k8s minion has a floating ip too? What is the floating ip of k8s minion used for?
I think, there is no reason.
Historically, Our heat template come from larsks/heat-kubernetes  template.
larsks/heat-kubernetes provides floating ip to minion nodes, this is just a reason.
The minion floating ips are needed to access the micro-service front ends from the internet.
-------------- next part --------------
An HTML attachment was scrubbed...
More information about the OpenStack-dev