Hi Philippe, I think the best option is the first one. In fact that is more or less the procedure established but FIWARE Lab handbooks. Only few notes: - Node name: you have to use SophiaAntipolis2 for the new node - Users: you have to use the same admin and services users for the new node - New node will be hidden in the cloud portal during federation tests - Old node will be accesible in the cloud portal till the new one is ready Anyway, Fernando (cc’ed) could tell you detailed information about the precess. BR -- Álvaro > El 19 Apr 2016, a las 19:27, Badia Philippe <philippe.badia at com4innov.com> escribió: > > Hello Alvaro, > We'd like to have your feedback concerning the upgrade of our node and the way we will move, from a Fiware lab keystone point of view, our current "Xifi" node to the upgraded one. > If more details or explanations are needed, we can set up call, let us know. > > Context: > Current configuration : Xifi Noderunning with 4 serveurs - OpenStack IceHouse > Targeted configuration : Fiware node running with 8 serveurs - OpenStack Kilo > > Here is an overview of the 2 approachs and the impact (in blue) for the Keystone Proxy integration. > > Solution 1 > > The goal is to minimise the impact of our upgrade for the current users, and want to keep the Xifi Node running as long as possible > We will deploy open stack on the new servers first. In this case we would need to configure the services endpoint as if it is a new node. > > - Deploy Openstack Kilo on the New servers > - Check Openstack services > - Request new service endpoints with Federation keystone ( Nova, Cinder, Neutron, Glance, Ceilometer) > Node name : Can we keep the same : SophiaAntipolis ? > Admin user : Can we keep the same admin-c4 ? > - Check connection with the Federation > - Snapshots existing VMs > - Stop the Xifi node and migrate the servers to the new node > - Migrate VMs for the Xifi infrastructure to the new node > How to handle the user impact on the cloud portal depending on the changes above ? > - Ficore node fully operational > > > Solution 2 > > We"ll keep the same service end point parameters for the upgraded node. > The users might be more impacted with this solution as the shutdown of the Xifi node will be done at the beginning of the plan. > > - Deploy Openstack Kilo on the 5 New servers > - Snapshot of existing VMs > - Stop the Xifi node and migrate the servers to the new node > - Check Openstack services > - Use the existing service endpoints with Federation keystone ( Nova, Cinder, Neutron, Glance, Ceilometer) > Node name remains unchanged : SophiaAntipolis > Admin user remains unchanged admin-c4 > - Check connection with the Federation > - Migrate VMs for the Xifi infrastructure to the new node. > How to handle the existing VMs re-creation on the new nodes ? > - Ficore node fully operational > > > Let us know your thoughts and what would be your preferred solution. > > Best regards, > Philippe > -- > Tel : +33 (0)4 89 86 69 37 > Mob: +33 (0)6 16 31 63 04 > www.com4innov.com <http://www.com4innov.com/> > > -------------- next part -------------- An HTML attachment was scrubbed... URL: <https://lists.fiware.org/private/fiware-lab-federation-nodes/attachments/20160421/1692975f/attachment.html>
You can get more information about our cookies and privacy policies clicking on the following links: Privacy policy Cookies policy