Stefan, Assuming this is not a disaster recovery (DR) scenario but a single data center:
Currently there is no mechanism to synchronize Solaris Cluster services across physically separate clusters. However, if you combine all of your tiers into a single cluster, then you can use resource groups and resource dependencies to synchronize the startup and shutdown across all the tiers. On 02/ 8/10 12:16 AM, Hartmut Streppel wrote: > Hi Stefan, > I doubt that you can apply the term "atomic step" to a switchover of a > set of services running on different clusters to a DR site. > > What you can do easily is add all of the switchover commands into one > big script, that you can run in the disaster case. There are problems > though, that are out of control of this solution. There might be start > order dependencies. You would have to implement the same mechanism > that you use to start your set of applications in the primary DC today. > > Regards > Hartmut > > On 02/04/10 17:11, Stefan Mueller-Wilken wrote: >> Dear all, >> >> I currently design a multi-tier server infrastructure with a primary >> and a secondary line of servers. >> >> The plan is to build an Open HA cluster for every tier (web, app, db, >> ...) with two servers each in fail-over configuration. >> While it is generally okay to switch each cluster pair autonomously >> (primary to backup web, primary to backup app ...), I need to switch >> over the complete set (primary web, app, db, ...) in one go when >> upgrading the software. >> >> Question now is: is there a way to orchestrate the clusters to switch >> over consistently, means, if to switch them all over to backup in one >> atomic step? And also: is it possible to do so automatically and >> based on heartbeat or agent intervention? >> >> Are resource groups the way to go? >> >> Thanks >> Mobi. >> >
