RAC DB patching - no downtime
I have a production RAC DB which running on 2 nodes with 11.2.0.3.0 release. Recently I have a request to patch the DB in rolling fashion with no downtime.
I understand that we can patch the 2 nodes RAC one after another (rolling fashion), shutdown 1st node's DB and patch while open 2nd node remain open for user access, after 1st node finished, bring up 1st node instance and do the same for 2nd node.
Does shutting down 1st node will really caused no downtime? what about the existing connection?
I know that user can configured to access via services, thus when 1st node down existing connections will auto switch to 2nd node, what about the transactions (DML) are running during the 1st node down? Would these sessions will be switch to the next available node without interuption?