Clustering with Solaris Zones

[trouphaz]

I found a whitepaper from Symantec that recommends putting zones on local storage in order to help with patching.  Here is a link to the doc, just in case. http://eval.symantec.com/mktginfo/enterprise/white_papers/ent-whitepaper_implementing_solaris_zones_...

I just want to make sure I'm grasping this and then ask a question because if I'm making a mistake here, my question might not make sense.

- If I have a 2 node cluster, each node will have its own local copy of the zone.

- The IP address is not configured in the zone config, but is instead controlled by VCS so that when patching the offline node, Solaris can boot the unused zone to patch it without having a network conflict.

- In essence, you have 2 distinct zones (one per server) that are configured the same so that VCS can boot either and apply the virtual IP and make it appear as if it were one.

My question is this.  If I'm using local disk for my zones and maintaining 2 distinct copies of this zone, how do I handle failover of my application between the two?  Say I have an Oracle database running in this zone.  How do I configure this app within the cluster such that I can start it on server 1 or move it to server 2 when I want to patch server 1?

[trouphaz]

I have found a possible answer, but I'm hoping someone else can verify that Symantec supports it.  You can mount and unmount filesystems within the global zone and make them available to the local zone while it is online.  If you have your zone mounted under /local/zone1, you can mount a filesystem as /local/zone1/root/newmount and it'll be seen within the zone as /newmount.  It also needs to be umounted from the global zone.  The only issue though is that once it is mounted under there, it is not visible to the global zone at all.

So, can VCS 5.0 manage mounting, umounting and monitoring these filesystems?  If so, they I can put the zone on local storage and the app on shared storage and then just mount the app on whichever zone is active at the tiem.