1Z0-078 Oracle Database 19c: RAC, ASM, and Grid Infrastructure Administration Practice Questions

This command correctly sets up the service. -preferred fm-node1,fm-node2 meets requirement 2. -available fm-node3 meets requirement 3. The combination of -commit_outcome TRUE and -failovertype TRANSACTION enables Application Continuity, which is required to transparently replay un-committed transactions (requirement 4). -policy AUTOMATIC ensures the service starts automatically with the database.

When using Oracle Managed Files with ASM, Oracle automatically handles the creation of unique filenames for new datafiles. During a PDB clone operation like this, you do not need to set PDB_FILE_NAME_CONVERT if the new files are intended to go into the same location (DB_CREATE_FILE_DEST). Oracle will generate new OMF names within that location for the new PDB's files.

Library cache objects (like SQL cursors and PL/SQL packages) must be consistent across all instances. GES manages the global locks to ensure this consistency.

When a transaction in one instance needs to lock a row that is also of interest to another instance, the TX enqueue must be managed globally by GES to prevent conflicts and ensure ACID properties.

To perform an OCR restore, the Oracle Clusterware stack must be stopped on all nodes. Then, on one node, you must start the stack in exclusive, no-CRS mode. This starts up the necessary low-level processes (like OHASD) to allow ASM access for the restore, but it prevents the full CRSD from starting, which would fail due to the missing OCR. Once in this mode, you can run ocrconfig -restore.

A failure group represents a set of disks that share a common point of failure (e.g., a power supply, a controller, or a storage shelf). To maximize availability, each component that could fail independently should be in its own failure group. Since each of the 6 disks is on a different shelf, each disk should be placed in its own failure group. This ensures that the mirrored copy of an extent on one disk is placed on a disk in a different failure group (and thus a different shelf).

The error message CRS-2527 explicitly states the server pool is not active on the target server. In a policy-managed cluster, resources are tied to server pools. If a server is not part of a pool's server list, the pool cannot be active there, and consequently, its member resources cannot start. The most direct resolution is to add the target node to the server pool's list of candidate servers using srvctl modify serverpool.

Hash partitioning distributes DML across different partitions and their underlying index segments, effectively spreading out the 'hot spots' and reducing contention on a single index leaf block.

Using a large sequence cache with NOORDER allows each instance to have its own range of sequence values. This prevents all instances from competing for and inserting into the same right-hand index leaf block, which is a classic cause of 'gc current block busy' waits.

Cache Fusion is the core technology of RAC that enables instances to share data without writing it to disk first. It uses the high-speed private interconnect to ship data blocks from one instance's buffer cache to another's, satisfying read or write requests much faster than a disk I/O operation would. This process is managed by the Global Cache Service (GCS).

This is a fundamental architectural requirement for Oracle RAC. Each instance manages its own transactions and requires its own UNDO tablespace. Because any instance may need to perform recovery on a transaction initiated by another (failed) instance, all UNDO tablespaces must be on shared storage accessible to all nodes in the cluster.

The key requirement is the recovery of in-flight transactions without user interruption. This is the precise function of Application Continuity (AC). AC records the session and transactional state, allowing it to be replayed on a surviving instance after a failure, making the outage transparent to the application. TAF is an older technology that primarily replays SELECT statements and does not recover the full transactional state.