I have a DW database running on 3 node 10.2.0.3 cluster . All the jobs (SELECT+DMLs ) uses service BATCH which is configured as load balanced (server side load balancing ) across 3 nodes. Wish to plan a rolling outage to apply oracle patches (Oracle support confirmed them as rolling patches ).
Here are the steps I’m thinking
1) Bring the listener on node 1 down (with service BATCH still running on node 1) , New connection now goes to node 2 and 3 only.
2) Let current sessions complete on node 1 and then bring down the node 1 for patching.
3) Apply the patch and bring up the node..
4) Repeat 1-3 for other nodes.
Am I missing any steps here?
I have a concern about Parallel Queries/DMLs ( The ones that gets kicked-off after listener is down ). Will they (PQ) go to the node 1 after bringing listener down? If so, how do I prevent them going to that node?
TIA for you responses.