On 7/21/14, 3:54 AM, BYEONG-GI KIM wrote: > Hello. > > I'm now trying to deploy two nodes environment (Flat Network or VLAN > Network) for OpenStack instead of 3 nodes which are generally considered > as default and recommended deployment for the OpenStack. The major > reason why I'm going to deploy such environment is performance issue. > > Anyway, here is a question; where should I put neutron-metadata-agent > in, i.e., Compute Node or Controller Node?
The neutron-metadata-agent must be run on the network node since all the metadata proxy processes that run in each router namespace must be able to communicate with it via a Unix domain socket. > As far as I know, the two node deployment has been supported by > Devstack, and the metadata agent is installed in Controller Node. In > this case, however, it seems that the meta data can not be High Available. > > I'd wanna deploy 2 nodes environment while inserting metadata-agent into > either Compute Node or Controller Node and achieving HA. I'm not sure how much HA you're going to get out of a two-node system if one is solely a compute node, you'd need two network nodes as well. Maybe you could describe how you were going to deploy things to achieve HA. -Brian _______________________________________________ Mailing list: http://lists.openstack.org/cgi-bin/mailman/listinfo/openstack Post to : [email protected] Unsubscribe : http://lists.openstack.org/cgi-bin/mailman/listinfo/openstack
