This mode is the default behavior of OPatchAuto. In rolling mode, Oracle services are shut down and patched one node at a time while the other nodes in the cluster remain fully operational. This method ensures continuous availability of the database throughout the patching process, as users can be failed over to active nodes. The prerequisites for starting a rolling session require the local node to be up and at least one remote node to be up and running. Rolling patching is the preferred method for most interim patches, as it incurs zero downtime. However, not all patches are eligible for rolling application; the patch's README file and metadata explicitly state if a patch can be applied in a rolling fashion.
The syntax itself, typically resembling opatchauto apply <PATCH_PATH> -oh <ORACLE_HOME> -nonrolling -exclusive , belies the complexity of the underlying operations. The utility must navigate the cessation of Cluster Ready Services (CRS), the modification of the Oracle Inventory, the patching of the Grid Infrastructure home, and finally the restart of the stack. It automates what would otherwise be a manual, error-prone sequence of dozens of commands. Yet, automation does not absolve the administrator of responsibility. Validating the patch success involves checking logs, verifying the inventory, and ensuring that the specific fixes contained within patch 72030 are active. opatchauto72030 execute in nonrolling mode exclusive
Running a rolling patch on a shared home will cause sudden, uncoordinated binary corruption and node evictions across active cluster hosts. This mode is the default behavior of OPatchAuto
# /u01/app/19.0.0/grid/OPatch/opatchauto apply /u01/software/p35319490 -nonrolling Use code with caution. Step 3: Handle Resumption (If Previous Attempt Failed) The prerequisites for starting a rolling session require