Oracle Database 19c: RAC, ASM, and Grid Infrastructure Administration Advanced Practice Exam: Hard Questions 2025

You are performing a rolling Grid Infrastructure patch on a 2-node RAC cluster. After patching node1, you run root.sh and it fails while configuring OHASD with errors indicating it cannot start oraagent.bin due to an "insufficient privileges"/"operation not permitted" message. The OS team recently enabled SELinux enforcing mode and tightened systemd hardening. Which action is the most appropriate to restore clusterware startup while maintaining a supportable configuration?

You are building a new 2-node RAC cluster with separate networks for public, private interconnect, and a dedicated storage network for iSCSI. The customer wants resilient name resolution even if corporate DNS is down. They also require that SCAN listeners remain available and correctly resolved by clients. Which design best meets the requirements with minimal operational risk?

A 3-node RAC cluster uses ASM with diskgroup DATA (NORMAL redundancy) across three failure groups (one per storage shelf). After a shelf failure, you replace the failed LUNs and add them back to ASM. You want to restore redundancy as quickly as possible but must avoid overwhelming the storage array during business hours. Which approach best balances speed and array impact?

Your ASM diskgroup FRA uses ASM Filter Driver (AFD). A new storage LUN is presented, but the sysadmin accidentally created an OS partition and formatted it with a filesystem before you could label it for AFD. Now, attempting to label it with 'asmcmd afd_label' fails. What is the best next step to make the LUN usable for ASM with AFD while minimizing risk of using the wrong device?

A RAC database uses ASM diskgroup DATA with a mix of SSD and HDD. The workload requires placing tempfiles and a specific high-IO tablespace on SSD, while other data remains on HDD. Which ASM feature and implementation provides the most appropriate control at the storage layer for this requirement?

A 4-node RAC database intermittently experiences global cache waits and spikes in 'gc cr block busy' and 'gc buffer busy acquire'. The issue correlates with one interconnect interface flapping on node3. Clusterware shows the private interface as up, but OS logs show frequent link resets. Which sequence is the most appropriate to confirm the root cause and mitigate impact with minimal downtime?

A service is configured for an OLTP application with TAF enabled. The application team reports that after an instance failure, sessions reconnect but in-flight transactions are lost and some sessions reconnect to an instance that is already overloaded. They want (1) transaction-level failover protection and (2) better load distribution on reconnection. Which solution best meets both goals using appropriate Oracle technologies?

After adding a new node to an existing RAC cluster, the node successfully joins and the database instance starts. However, services do not start on the new node even though the service is configured as 'AVAILABLE' on all instances. 'srvctl status service' shows the service running on old nodes only. No errors appear in alert logs. What is the most likely cause and the correct fix?

A 2-node RAC cluster shows frequent node evictions. In CRS logs you see messages indicating missed network heartbeats and a subsequent reboot of the evicted node. The private interconnect uses bonded NICs and seems stable under normal checks. Voting disks are in ASM on shared storage. Which advanced diagnostic approach most directly helps determine whether the root cause is network heartbeat loss vs. I/O latency to voting files?

A planned storage maintenance causes temporary path failures to ASM disks. ASM diskgroups remain mounted, but one RAC instance intermittently terminates with ORA-15064/ORA-27061-style errors during heavy I/O. You need to harden the configuration so brief path outages do not crash instances, assuming multipathing is correctly configured. Which change most directly improves tolerance to transient ASM disk I/O interruptions at the cluster/storage integration layer?