Our stand by is 3 node cluster ( 22.214.171.124 db on red hat 6.3). We want to patch one node at a time.
My mrp process is running on node 1.
How can i start my mrp process on node 2 ?
Will there be a downtime / delay ( for recover to happen) during which mrp process is switched from node 1 to node 2?
Recovery can be running at only one instance in a RAC DG environment. That instance must have access to all the archivelogs from the primary.
My guess is that you would need to
1. Stop (CANCEL) recovery on node 1
2. Shutdown the instance on node 1
3. Startup the listener on node 2
4. Startup mount the instance on node 2
5. Check that the listener and password file on node 2 are configured and the tnsnames on the primary is configured for the net name for the instance on node 2
6. set log_archive_dest_state_2 (or whichever you are using) on the primary to DEFER
7. configure log_archive_dest_2 (or whichever you are using) on the primary to the net name for the instance on node 2
8. set log_archive_dest_state_2 (or whichever you are using) on the primary to ENABLE
But I have never done so.
You should check with Oracle Support if they have a proper note on switching the MRP to another node in RAC DG.