It stops the Oracle Clusterware stack on the local node ( crsctl stop crs ).
Before executing any exclusive patching operations, verify your current OPatch and OPatchAuto versions. Outdated patching utilities frequently misinterpret patch metadata. Log in as the grid user and check the OPatch version: $ORACLE_HOME/OPatch/opatch version Use code with caution.
Do not attempt to patch Node 2 until Node 1 has fully completed, brought its cluster services back online, and exited clean.
In modern Oracle environments (12c and later), "non-rolling" doesn't mean everything is offline; rather, it follows a specific three-phase orchestration: Oracle Help Center Phase 1 (Local Node):
If your patching session aborts with the 72030 error, follow this structured, production-tested workflow to resolve the issue. 1. Verify and Clean Up Abandoned Locks
The most frequent cause of this error is a stale lock file left behind by a previous aborted or crashed opatchauto attempt.
Before running opatchauto72030 in non-rolling mode, thorough preparation is essential to ensure a smooth process.
By default, opatchauto applies patches in , which allows nodes to be patched one by one without total cluster downtime. However, certain patches or configurations (such as shared CRS homes ) are not "rollable". When opatchauto detects this, it throws the following message:
When this error triggers, it effectively acts as a safety mechanism to prevent system corruption that would occur if you tried to patch a shared home while other nodes were still using it. Detection Accuracy:
: For large clusters (8+ nodes), avoid running manual sequential opatchauto commands. Instead, patch a single software home in a sandbox, create a golden image, and use Oracle Fleet Patching and Provisioning (FPP) to push the updated software home out in mass non-rolling waves.
