The scenario is now I have configured RGOffload (affinity is not available for my release) to offload the oracle-dev-rg when oracle-prod-rg wants to start on the dev node. Since the oracle-dev-rg only has one node (dev) in its node list, according to the docs this should force it to shutdown since the oracle-prod-rg is the critical resource.
I am scheduling a test window to verify and if there are any problems I'll let everyone know. Thanks for the help. Norm On Mon, 2008-14-01 at 21:49 +0100, Thorsten Frueauf wrote: > Hi Norm et al, > > How did you configure the RGOffload resource? Is > continue_to_offload=true? See > http://docs.sun.com/app/docs/doc/819-0703/cdcigche?a=view for more details. > > I have no practical experience with RGOffload, so maybe others have an idea. > > The question is why did the move fail in the first place? > > Which SC 3.1 version are you using? Would it be an option to use > RG_affinities instead like described in > http://docs.sun.com/app/docs/doc/819-1405/x-1dill?a=view in the > "SUNW.RGOffload" section? > > If the RG to be offlined has only one node in its Nodelist, it would > turned just off instead of getting relocated. > > Hope this helps. > > Greets > Thorsten > > Norm Dressler wrote: > > Is there a way to just shutdown a resource rather then have it move to > > another server? > > > > I have a 2 node cluster. If production fails on node 1, I want it to > > move to node 2 but have dev running there shutdown. The RGOffload tries > > to move the resource and if it fails, it restarts on the originating > > node. I just want it to shutdown - > > > > Would that be the affect if I disabled the failover option for resource > > group? > > > > Norm > > > > On Fri, 2007-14-12 at 14:46 +0100, Thorsten Frueauf wrote: > >> Hi Norm, > >> > >> I am assuming that you have at least two resource groups configured - > >> one for the production resources, and one (or even two) for your test > >> and development resources. > >> > >> If so, then you can use the strong negative affinity in order to offload > >> the test/dev resource group, before it is getting started on the > >> secondary node. > >> > >> The RG_affinities property for resource groups got introduced with SC > >> 3.1 U3. > >> > >> If you are running an earlier update of SC 3.1 you may use the RGOffload > >> resource type (obsolted by affinities). > >> > >> See > >> http://docs.sun.com/app/docs/doc/819-2974/ch14_resources_admin-42?a=view > >> for more details on affinities. > >> > >> See > >> http://docs.sun.com/app/docs/doc/819-0582/6n30g7ilc?l=en&a=view&q=RGOffload > >> > >> to see the manpage for SUNW.RGOffload. > >> > >> Greets > >> Thorsten > >> > >> Norm Dressler wrote: > >>> I have a 2 node cluster running various Oracle schemas. > >>> > >>> With a recent upgrade, we have determined we can no longer run three of > >>> our environments on the backup server. > >>> > >>> Example, primary server runs production, secondary runs test and dev. > >>> > >>> In a failover situation, production should failover to the secondary. > >>> However, the secondary does not have sufficient resources to allow it to > >>> start. > >>> > >>> The solution is to shut down test prior to failing over. This must > >>> happen before Oracle tries to start on the secondary server. The scripts > >>> used to start Oracle are the same across the environments and the only > >>> difference is the SID. > >>> > >>> We also use data storage resources and it has the the ability to set the > >>> RunBeforeStartMethod parameter and since it is one of the first items to > >>> fail over, I thought I found my solution but unfortunately it did not run > >>> the script I wrote to shutdown test - the script consisted of scswitch > >>> commands. > >>> > >>> Does anyone know of a way to stop a resource on the secondary server > >>> prior to a resource failing over to it? > >>> > >>> Thanks for any assistance. > >>> > >>> Norm > >
