Ok, I have shot myself in the foot. I missed SSO in the detailed design of a (would be simple) upgrade of vSphere 4.1 to 5.1.
Installed two vCenters 5.1, witht the requirement for linked mode, so Installed SSO DB on VC1 (including SQL2008 on the VC itself for a small environment) and linked the VC2 to that.
All fine, next comes in SRM with NetApp replicated storage. SRM testing etc works fine, failover also works ok for a selected set of machines.
Now customer wants to do a DR test by shutting down main site... where VC1 lives.. But in that case SSO DB is not reachable, so I am not able to start SRM, right?
If I read all the so fine VMware SSO docs, it is not possible to upgrade an existing SSO installation to a clustered SSO. Shoot.
Is it possible to create a a new set of VM's with SSO and re-import the SSO data from the other SSO instance into the clustered SSODB?
So I "remove" the SSO instance from VC1 and create a HA clustered SSO to point VC1 and VC2 to?
Or am I really FUBAR and do I need to start from scratch for both sites? Including all fine third party plugins etc.. (which can be quite a few in the end).
If I read this blogpost: Linked mode with SSO for SRM | VMware vSphere Blog - VMware Blogs
Linked mode vCenter instances need to authenticate against what appears to be a single SSO instance, whether it is a single SSO instance at one site, multiple SSO instances in HA mode, or multiple SSO instances in multi-site mode. Of those options when using SRM, my preference is to either not use linked mode at all, or to deploy SSO in multisite mode. A single instance or an HA cluster at one site will always introduce the risk that you can not log in during a disaster.
So what are my options here? Can a upgrade to SSO 5.5 help me in any way?
Thanks for you wisdom...
Marcel
