01. Oracle Support needs access to the schemas of an Autonomous AI Database Serverless instance for a two-day investigation. The administrator runs the call below, and it is rejected.
EXEC DBMS_CLOUD_ADMIN.ENABLE_OPERATOR_ACCESS(auth_duration => 48);
Why is the call rejected?
a) Operator access can be granted only while the instance is stopped
b) The duration is given in minutes and must be a multiple of 60
c) Only Dedicated deployments support operator access through this call
d) The duration is a whole number of hours from 1 to 24
02. A development team has written a lookup routine as a JavaScript function that runs in the database with Multilingual Engine (MLE). They want agents connected through the Autonomous AI Database MCP Server to call this routine as a tool.
What should the team do?
a) Publish the JavaScript function through ORDS, which then lists it as a tool
b) Rewrite the routine in PL/SQL, which may call Java stored procedures, and register it
c) Register the JavaScript function directly as the function attribute of the tool
d) Wrap the function in an MLE module and grant EXECUTE on it to the MCP login user
03. An operations group can view the Autonomous AI Databases on Dedicated Exadata Infrastructure in compartment OpsComp through read autonomous-databases. The group must now also be able to take manual backups of those databases, and nothing else about its access should change.
Which statement should be added?
a) Allow group OpsGroup to read autonomous-backups in compartment OpsComp
b) Allow group OpsGroup to use autonomous-container-databases in compartment OpsComp
c) Allow group OpsGroup to manage autonomous-backups in compartment OpsComp
d) Allow group OpsGroup to manage autonomous-databases in compartment OpsComp
04. On an Autonomous AI Lakehouse database that uses the predefined resource plan, ad hoc queries submitted through the HIGH service sometimes run for hours. The DBA wants any statement on HIGH that runs longer than 30 minutes to be cancelled automatically, without affecting the other services.
Which call achieves this?
a) CS_RESOURCE_MANAGER.UPDATE_PLAN_DIRECTIVE(consumer_group => 'HIGH', elapsed_time_limit => 30)
b) CS_RESOURCE_MANAGER.UPDATE_PLAN_DIRECTIVE(consumer_group => 'HIGH', io_megabytes_limit => 1800)
c) CS_RESOURCE_MANAGER.UPDATE_PLAN_DIRECTIVE(consumer_group => 'MEDIUM', elapsed_time_limit => 1800)
d) CS_RESOURCE_MANAGER.UPDATE_PLAN_DIRECTIVE(consumer_group => 'HIGH', elapsed_time_limit => 1800)
05. After obtaining a token, an engineer configures an MCP client with the following server URL and adds the token to the client configuration:
https://dataaccess.adb.us-ashburn-1.oraclecloudapps.com/adb/auth/v1/databases/ocid1.autonomousdatabase.oc1..xyz/token
Every connection attempt returns HTTP 404. What is the cause?
a) The database user lacks privileges on the tools that the client tries to list
b) The token has expired, since tokens from this endpoint stay valid for only a short time
c) The URL is the token endpoint rather than the /adb/mcp/v1/databases/ endpoint
d) The token has to be passed as a query parameter of this URL instead
06. Which two statements about Autonomous Data Guard on Dedicated Exadata Infrastructure are true?
(Choose two.)
a) Protection mode is fixed at Maximum Availability for every configuration
b) The standby ACDs can be placed in as many as four different regions
c) In Maximum Performance mode, the failover lag limit defaults to 30 seconds
d) After a failover, the former primary is a disabled standby until reinstated
e) Each database in an ACD can be given its own standby in a chosen region
07. An Autonomous AI Database must query Apache Iceberg tables that are registered in Databricks Unity Catalog through its Iceberg REST interface. The data engineer planned to rely on the catalog to vend temporary object-store credentials to the database.
Which statement about this plan is true?
a) Unity does not vend credentials here, so the database needs its own object-store credential
b) The tables must first be copied into the database, since Iceberg data is not queried in place
c) Unity is not a supported catalog, so the tables must be registered in AWS Glue instead
d) The tables must be converted to Delta format before a Unity catalog can serve them
08. A support team keeps several hundred product manuals as PDF files in an OCI Object Storage bucket, and new manuals are added each week. Agents want to ask questions in natural language and receive answers drawn from the manuals rather than from the LLM's general training.
Which approach meets this requirement with Select AI?
a) Use the chat action with a profile that lists the bucket's URI in its object list
b) Create an external table over the bucket, grant it to the agents, and ask with runsql
c) Load the manuals into a CLOB column and ask with showsql to return the passages
d) Index the bucket with a Select AI vector index, add it to a profile, and use narrate
09. Each night a large file is loaded into a table of an Autonomous AI Lakehouse database with SQL*Loader in direct-path mode. Next-morning reports on that table get poor execution plans, and its optimizer statistics turn out to be stale.
What should the team do?
a) Set OPTIMIZER_IGNORE_HINTS to FALSE so that the optimizer rereads the table
b) Enable automatic indexing, which also refreshes statistics on loaded tables
c) Gather the table's statistics after each load, since this load path does not
d) Run the load through the HIGH service, which gathers statistics as it loads
10. A company must share a curated sales dataset with an external partner whose analysts use Apache Spark and pandas, not Oracle Database. The partner should receive consistent versions of the data that the provider publishes.
Which approach fits?
a) A versioned Data Share over the Delta Sharing protocol
b) A Cloud Link registered with a tenancy-wide scope
c) A Live Share between provider and partner databases
d) A database link that the partner creates to the provider