Since there are a million ways to deploy OpenStack, I guess it's hard to provide detailed upgrade instructions that fit all cases.
The operations guide has a somewhat high-level chapter on upgrading. To that, you should add the upgrade sections of the relevant release notes, for example Nova/Ocata (scroll down to the Upgrade Notes section).
There are also instructive summit presentations on upgrades, such as this recent one.
Most, perhaps all core services allow rolling upgrades with no or very minimal downtime. This means that your cloud runs versions n and n+1 in parallel during the upgrade.
Asked: 2017-06-20 03:42:15 -0600
Seen: 68 times
Last updated: Jun 20 '17
Quite open question, there are many methods, all depend on your current deployment, available resources, downtime allowed, etc. It differs too much from a blue-green upgrade to a live upgrade to an offline upgrade. Please, provide more information about your current deployment and your goals.
Also, what services are you using, they have different upgrade methods, what openstack release and what ha solution is implemented (keepalived/haproxy, pacemaker/corosync)
@Eduardo Gonzalez i am looking out for upgrade solutions with minimal downtime for an internal production cloud running newton.(no shutdown) For the basic services like neutron, nova, glance, horizon, heat, cinder, keystone with haproxy i am just trying to identity procedures to go about it
@Eduardo Gonzalez i basically want to understand the starting point to begin with the upgrade so that i will be able to design solution which would help me upgrade to higher versions of openstack in future