Sounds reasonable. Out of curiosity what people use or do in order to rollback a failed upgrade or an upgrade that can bring instability in the control or operation of the cluster ? In VMware as vCenter used to run within a Virtual Machine I used to take a snapshot before starting the upgrade. In oVirt if I have my Engine running in a simple VM somewhere else out of the cluster as well I guess that could also be an option. What about when it is a self hosted engine. Would it work ? Thanks Fernando On 24/01/2017 05:03, Yedidyah Bar David wrote:
On Mon, Jan 23, 2017 at 9:36 PM, Fernando Frediani <fernando.frediani@upx.com.br> wrote:
I am deploying a oVirt environment which will not get production data immediately.
Obviously I would rather use 4.1 RC due the many changes and fixes present. Later when 4.1 becomes stable then upgrade to it. Does anyone see any problem in doing that way or would it be more advisable to start with 4.0.6 and upgrade to 4.1 stable when time comes ? If eventually it will be production, I'd start with 4.0.6.
Generally speaking, if we find in the future a bug when upgrading from 4.0.6 to 4.1.z, we'll try to solve it, so that it does not affect an upgrade from 4.0.6 to 4.1.z+1. But if we find a bug that affects only upgrade from a rc/beta/etc. version to a stable version, we might decide it's not worth fixing.
Also note that 4.1.0 should be out really soon:
https://www.ovirt.org/develop/release-management/releases/4.1/release-manage...
So you might as well simply wait a bit.
My concern are the issue people related they had when upgrading from one major version to another in the past. In general, or to/from beta/rc/etc versions?
In general we appreciate very much people testing upgrades from/to beta/rc versions, and if they find bugs, we do try to fix them. But people should do this in test environments, not ones that are eventually destined to become production.
Best,