Conversion issues and patterns reference

Conversion workspaces aggregate all conversion issues into groups and categories to help you plan for fixing conversion errors and warnings. Each category represents the type of work you might need to perform to fix the issues (review, refactor, adjust data types). Groups provide further aggregation, as they differentiate between specific cases at a lower level:

Conversion workspace overview screen that shows conversion issue groups and categories.
Figure 1. Conversion workspace overview screen with conversion issue groups and categories.
Conversion workspace overview screen that shows conversion issue groups and categories.

The following table lists all conversion issue groups you can encounter during schema conversion:

Issue group ID Description

Invalid source code

Potential root cause

Errors in this group often occur when Database Migration Service encounters unknown syntax, or when the Oracle source code isn't valid (for example, a stored procedure is missing the END; keyword.)

Possible mitigation

Correct the invalid objects in the source Oracle database. Then, refresh the source schema snapshot in Database Migration Service and re-try the schema conversion process. Alternatively, you can exclude the object from the migration.

Missing referenced objects

Potential root cause

Database Migration Service uses metadata of objects in the source tree to improve the code conversion quality of dependent objects. If your code refers to objects that aren't included in the source schema, you might experience conversion issues because Database Migration Service can't determine the structure or data types for the missing referenced columns, attributes, or objects.

Possible errors in this group include incorrect data types for user-defined types (UDTs) or default VARCHAR data types for columns, parameters, or variables.

Possible mitigation

Either ensure that all referenced objects are added to the Database Migration Service source tree, or manually adjust the PostgreSQL code based on your knowledge of the source data model for the missing dependencies.

Tables without primary key

Potential root cause

Database Migration Service requires all tables to have a primary key. For tables without primary keys, Database Migration Service adds a NUMERIC column named rowid to the target PostgreSQL table. This column is populated with corresponding numeric values from the source Oracle ROWID pseudocolumn. To ensure that INSERT statements don't fail after the migration, Database Migration Service creates a sequence and uses it to auto-increment the rowid column.

Possible mitigation

You can either retain or remove the rowid column after the migration.

Unsupported Oracle built-in functionality

Potential root cause

You might be using built-in Oracle functionality that isn't supported.

Possible mitigation

Find similar functionality in PostgreSQL and modify the converted code accordingly. In some cases, the missing functionality might be provided by the Orafce extension, available for both Cloud SQL for PostgreSQL and AlloyDB for PostgreSQL migrations.

SQLCODE not yet supported

Potential root cause

The Oracle SQLCODE function isn't supported for conversion. SQLCODE returns an INTEGER, whereas the closest equivalent in PostgreSQL is the SQLSTATE function that returns TEXT values.

Possible mitigation

If your source code doesn't rely on SQLCODE returning an integer (for example, SQLCODE is only ever logged in a VARCHAR2 column or with a DBMS_OUTPUT message), then SQLSTATE can be safely used in the PostgreSQL code.

If your source code relies on SQLCODE returning an integer (for example, you use it with NUMBER or INTEGER variables, or save SQLCODE return values as parameters or columns), you need to refactor the converted code.

Oracle SQL function not yet supported

Potential root cause

Some Oracle built-in functions are unsupported by Database Migration Service for conversion.

Certain functions might have their equivalents in PostgreSQL (for example, ASCII), others might share the same name but have different specifications (for example, REGEXP% functions). Some functions might not exist at all.

Possible mitigation

Inspect which Oracle function raised the error.

  • If the function exists with the same behavior in PostgreSQL, you can disregard the error message as the converted code should work the same after the migration.

  • If a function has the same name but works differently or just doesn't exist in PostgreSQL, you can try to fix the converted code:

    • Create a User-Defined Function (UDF) of the same name in the destination database.
    • Replace the function call in the source with an equivalent expression.

    • Use Gemini-enhanced conversion assistance. For more information, see Convert SQL Server code and schema with Gemini assistance.

    • Some Oracle database functions or features may be replicated with extensions, such as Orafce. For more information on extensions supported in your destination database, see Supported database extensions.

Oracle built-in packages not fully supported

Potential root cause

Database Migration Service supports certain Oracle built-in packages, but many don't have full conversion support, for example DBMS_STATS, DBMS_UTILITY, or DBMS_SQL.

Possible mitigation

If you use any unsupported packages, you might need to:

  • Change your code. For example, instead of using DBMS_STATS.GATHER_TABLE_STATS, you might use a simpler command such as ANALYZE.

    Packages such as UTL_FILE and DBMS_AQ might require refactoring because they can't be easily replicated in PostgreSQL.

  • Some Oracle database functions or features may be replicated with extensions, such as Orafce. For more information on extensions supported in your destination database, see Supported database extensions.

Oracle data type not yet supported for conversion

Potential root cause

Some Oracle data types aren't currently supported for conversion or data movement.

Possible mitigation

In most cases, PostgreSQL has an equivalent data type. You can use conversion mapping files to customize conversion logic and transform the unsupported Oracle data type to the required PostgreSQL data type.

For more information on data type support, see Conversion workspaces - overview and supported objects.

Source feature not yet supported
Potential root cause

This group captures all generic issues that relate to Oracle features that aren't supported for conversion. Issues in this group don't fall under any other, more specific, issue groups.

Possible mitigation

Schema object type not supported

Potential root cause

Database Migration Service doesn't support certain Oracle schema object types for code conversion because PostgreSQL doesn't have appropriate equivalents. Examples include Index-Organized Tables (IOTs), text search indexes, or bodies for user-defined types (UDTs).

Possible mitigation

Database Migration Service converts unsupported objects into the closest PostgreSQL equivalent. For example, IOTs become regular tables with a primary key constraint, text search indexes convert into B-tree indexes. Keep in mind that these conversions might lead to loss of functionality specific to the original object type.

PL/SQL feature not yet supported
Potential root cause

This group captures all generic issues that relate to PL/SQL features that aren't supported for conversion. Issues in this group don't fall under any other, more specific, issue groups.

Possible mitigation

Bulk binding not yet supported

Potential root cause

Database Migration Service code conversion doesn't currently support Oracle bulk binding features such as BULK COLLECT, FORALL or SAVE EXCEPTIONS.

Possible mitigation

To fix the issue, you need to modify the code that uses bulk binding features. You might want to consider the differences in PostgreSQL and Oracle architecture and whether array processing is necessary in PostgreSQL for your use case.

There are several strategies for approaching Oracle bulk binding operations in PostgreSQL. Their usage depends on your specific scenario, so we recommend that you use Gemini-powered conversion assistance to handle your specific needs. Here are some other example recommendations to help get you started:

  • For a full BULK COLLECT (no LIMIT), you can try to use the ARRAY_AGG function.
  • For BULK COLLECT with LIMIT, you can try to use a CURSOR FOR LOOP to load and process batches of rows in arrays. But, if your case allows for functional changes, a possibly simpler alternative would be to use the CURSOR FOR LOOP to process one row at a time (rather than load them into arrays).
  • For FORALL operations, you can try set-based DML with UNNEST if you choose to use array processing.
  • For SAVE EXCEPTIONS, you might need to write an exception handler within a row-based CURSOR FOR LOOP as there is no equivalent clause in PostgreSQL.

Collections not yet supported

Potential root cause

Database Migration Service code conversion has partial support for Oracle collections.

Possible mitigation

You need to modify the converted PostgreSQL code accordingly. When resolving issues with collections, remember that PostgreSQL arrays are never sparse. If you assign elements sparsely, PostgreSQL arrays might return different results and cardinality counts than Oracle arrays.

Because PostgreSQL doesn't support string-indexed arrays, depending on the nature of the data, you might find JSON/JSONB or the hstore extensions to be suitable.

Pipelined functions not yet supported

Potential root cause

Pipelined functions are not supported by Database Migration Service.

Possible mitigation

You can replace Oracle pipelined functions with PostgreSQL set returning functions. We recommend that you adjust the code in a way that is relevant to your use-case. Here are some examples to help get you started:

  1. Reference the PostgreSQL type (UDT) that is converted from the source Oracle object or record type that defines the pipelined function's rowtype. Then, modify the PostgreSQL function's return clause to be RETURNS SETOF <type name> or RETURNS TABLE, depending on your specific case.

  2. Replace the return value in the converted PIPE ROW code with RETURN NEXT <row or record variable>.

The source pipelined function's collection type, used in the Oracle function's RETURN clause, isn't needed in PostgreSQL.

Dynamic SQL option not yet supported

Potential root cause

Database Migration Service provides partial support for converting dynamic SQL. Oracle EXECUTE IMMEDIATE, OPEN FOR, and USING/INTO keywords are correctly converted to their PostgreSQL equivalents, but the actual dynamic SQL, DML, or DDL strings aren't converted.

Possible mitigation

You need to modify the converted code to match your requirements. We highly recommend that you use Gemini-powered conversion assistance to handle dynamic SQL.

CONNECT BY option not yet supported

Potential root cause

Most CONNECT BY operators, functions, and pseudocolumns are supported by Database Migration Service and converted to PostgreSQL recursive Common Table Expressions (CTEs). But there are certain exceptions that might require your attention. For example, the ORDER SIBLINGS BY clause is supported for ascending order only.

Possible mitigation

It's not possible to replicate the ORDER SIBLINGS BY clause for descending order in a simple manner, so you might need to re-structure your code to work with ascending order.

Review the reported issues with CONNECT BY operators, and adjust the code where necessary. We recommend that you explore Gemini-powered auto-conversion features for expediting such fixes. For more information, see Convert Oracle code and schema with Gemini assistance.

JSON not yet supported

Potential root cause

There are certain limitations in how Database Migration Service supports JSON for data movement and code conversion:

  • JSONB isn't supported for data movement.
  • Code conversion doesn't support the JSON query functions or operators in Oracle (such as JSON_TABLE, JSON_QUERY, JSON_OBJECT[AGG], JSON_ARRAYAGG, JSON_PATCH).
  • In Oracle versions earlier than 21c, you can store JSON data in VARCHAR2, CLOB, or BLOB columns, and verify it with the IS JSON condition. Database Migration Service doesn't support converting this condition.
Possible mitigation

  • To move JSON data stored in VARCHAR2, CLOB, or BLOB columns to PostgreSQL, you might need to write a conversion mapping file with the MODIFY_TYPE directive.

    For example:

    MODIFY_TYPE SOURCE_TABLE_NAME:BLOB_COLUMN_NAME_WITH_JSON_DATA:JSON

    For more information on conversion mapping files, see Conversion mapping files.

  • To convert the Oracle JSON functions and operators to PostgreSQL, you can use PostgreSQL functions such as JSONB_ARRAY_ELEMENTS, JSON_AGG. For more information, see JSON Functions and Operators in the PostgreSQL documentation.

Locking and transactions issues

Potential root cause

Database Migration Service code conversion doesn't support LOCK or SAVEPOINT statements. PostgreSQL doesn't support SAVEPOINT.

Possible mitigation
  • Break the SAVEPOINT blocks in Oracle to separate transaction blocks in PostgreSQL that utilize the ROLLBACK statement.
  • To implement DBMS_LOCK, use the PostgreSQL pg_dbms_lock extension. This extension simplifies the migration of Oracle user-defined locks to PostgreSQL by emulating the functionality of the Oracle DBMS_LOCK package.

XML not yet supported

Potential root cause

Database Migration Service doesn't support Oracle XMLTYPE or the associated XML functions and operators.

Possible mitigation

Although Database Migration Service doesn't directly support XMLTYPE, you can customize BLOB, CLOB, NVARCHAR2, or VARCHAR2 columns to XML for data movement. PostgreSQL supports XML functionality.

To migrate your XML data, follow these steps:

  1. Create a conversion mapping file with the MODIFY_TYPE directive for XML data. For example: 

    MODIFY_TYPE SOURCE_TABLE_NAME:BLOB_COLUMN_NAME_WITH_XML_DATA:XML

    For more information on conversion mapping files, see Conversion mapping files.

  2. Start the migration job. This process migrates all data from Oracle to PostgreSQL, except for data in columns of type XMLTYPE. These columns are filled with NULL values in PostgreSQL.t
  3. Create a foreign table in PostgreSQL by selecting only the XMLTYPE column from Oracle. Include the primary key column from the source table.
  4. Merge the XML data from the foreign table into the original PostgreSQL table.

For more information on how PostgreSQL works with XMLTYPE, see XML Functions in the PostgreSQL documentation.

PIVOT not yet supported

Potential root cause

Database Migration Service doesn't support the PIVOT and UNPIVOT transpose operators for code conversion.
Possible mitigation

You can achieve PIVOT transposition in PostgreSQL by using the tablefunc extension or by rewriting the source PIVOT expression to conditional aggregations. For example:

  • SUM(CASE WHEN <condition> THEN <value> ELSE 0 END)
  • COUNT(CASE WHEN <condition> THEN 1 END)

You can create UNPIVOT transpositions, which create key-value pairs from a set of columns, in PostgreSQL in a variety of ways:

  • Combining a LATERAL join with a set of VALUES. For example: SELECT ... FROM <table> CROSS JOIN LATERAL (VALUES ('<column1-name>', <column1>), ('<column2-name>', <column2>) AS u(column_name, column_value))
  • Using UNNEST with ARRAYs. For example: SELECT ..., UNNEST(ARRAY['<column1-name>', '<column2-name>']) AS column_name, UNNEST(ARRAY[column1, column2]) AS column_value FROM <table>

ALTERstatement option not yet supported

Potential root cause

Database Migration Service doesn't ALTER statements (that are often executed in dynamic SQL) for conversion.

Possible mitigation

Replace the Oracle ALTER statements with the SET command in PostgreSQL. We recommend that you explore Gemini-powered auto-conversion features for expediting such fixes. For more information, see Convert Oracle code and schema with Gemini assistance.

SQL feature not yet supported
Potential root cause

This group captures all generic issues that relate to SQL features that aren't supported for conversion. Issues in this group don't fall under any other, more specific, issue groups. Examples include database event triggers, GRANT statements, multi-table INSERT operations, DML on inline views (INSERT INTO (SELECT ... FROM ...)), lateral views.

Possible mitigation

Unsupported syntax
Potential root cause

This group captures all generic issues that relate to unsupported Oracle SQL or PL/SQL syntax. Issues in this group don't fall under any other, more specific, issue groups.

Possible mitigation

Change your code to use PostgreSQL functionally-equivalent PostgreSQL syntax. We recommend that you explore Gemini-powered auto-conversion features to adjust the code. For more information, see Convert Oracle code and schema with Gemini assistance.

Unsupported SQL syntax

Potential root cause

Your source code uses SQL syntax or elements that are unsupported by Database Migration Service. For example, the NLS_LANG parameter in the Oracle TO_DATE function is not supported.

Possible mitigation
Unsupported PL/SQL syntax

Potential root cause

Your source code uses PL/SQL syntax or elements that are unsupported by Database Migration Service. For example, record-based INSERT statements (such as INSERT INTO table VALUES r_variable) and PRAGMA RESTRICT_REFERENCES are unsupported.

Possible mitigation

Change your code to use PostgreSQL functionally-equivalent PostgreSQL syntax. We recommend that you explore Gemini-powered auto-conversion features to adjust the code. For more information, see Convert Oracle code and schema with Gemini assistance.

Unsupported date and timestamp syntax

Potential root cause

Database Migration Service might raise errors or warnings for unsupported date or timestamp syntax, operations, or expressions. Examples of these issues include comparisons between mismatched data types or the use of TZH:TZM format model in timestamps. For more information on supported objects and data types, see About conversion workspaces.

Possible mitigation

You can re-create most date and timestamp expressions using PostgreSQL equivalents. We recommend that you explore Gemini-powered auto-conversion features for expediting such fixes. For more information, see Convert Oracle code and schema with Gemini assistance.

Unsupported elements of Oracle's exception handling syntax

Potential root cause

Database Migration Service code conversion doesn't support the following Oracle PL/SQL exception syntax elements:

  • The use of RAISE_APPLICATION_ERROR with variable error codes instead of literal -20nnn codes.
  • The use of SQLERRM with an error code argument, as this syntax is unsupported in PostgreSQL.
Possible mitigation

You must manually resolve these issues in the converted code. We recommend that you explore Gemini-powered auto-conversion features for expediting such fixes. For more information, see Convert Oracle code and schema with Gemini assistance.

Data types and conversion issues

Potential root cause

Database Migration Service can group conversion issues based on the context (for example, conversion issues that occur in type comparison expressions). The CW_OP0500 group captures all generic data type conversion problems that don't fall under other issue groups.

Possible mitigation

In most cases, Database Migration Service emits an ERROR_UNIMPLEMENTED or ERROR_TYPE message in the converted PostgreSQL code. Resolve this error based on your knowledge of the data types involved.

Date format mask issues

Potential root cause

You might encounter warnings or issues when converting date or timestamp expressions to or from strings based on a format model. Database Migration Service uses a default model (currently DD-MON-RR) when casts in the Oracle source code exclude an explicit date or timestamp format model.

This can sometimes cause issues in the converted code if the format model emitted for the implicit cast conflicts with an explicit format model in the same expression. You might also see this issue if your data is likely to be affected by the differences between Oracle RR format and PostgreSQL YY format.

Possible mitigation

Review and validate the converted PostgreSQL expressions in the conversion workspace.

Numeric format mask issues

Potential root cause

Database Migration Service doesn't support all Oracle format models. For example 'L' or 'X' aren't supported. You might encounter issues or warnings with code that converts strings to numbers based on Oracle format models.

Possible mitigation

For Oracle format models that have no equivalent in PostgreSQL, you might need to refactor your expressions or format models. We recommend that you explore Gemini-powered auto-conversion features for expediting such fixes. For more information, see Convert Oracle code and schema with Gemini assistance.

Data type casting issues

Potential root cause

You might encounter errors due to unsupported or inaccurate data type casting. Database Migration Service usually emits ERROR_UNIMPLEMENTED. This is usually because of missing or incomplete metadata. Database Migration Service might not have enough information to cast a type, for example in function arguments or procedure parameters.

Possible mitigation

Adjust the PostgreSQL code to ensure correct data type conversions. These corrections require that you are familiar with your referenced attributes, variables, and columns.

Comparison issues

Potential root cause

Database Migration Service might not have enough metadata or information about data types when converting data comparison expressions. For example, this can happen when a user-defined type (UDT) is compared with NULL.

Possible mitigation

Review the converted PostgreSQL expressions and resolve the issues. We recommend that you explore Gemini-powered auto-conversion features for expediting such fixes. For more information, see Convert Oracle code and schema with Gemini assistance.

Issues in this category represent cases where Oracle source code is correctly converted to the closest PostgreSQL equivalent, but the resulting code might have minor semantic or functional differences that require your review. This happens because of the differences in how Oracle and PostgreSQL handle data types, formats, or objects.

At first glance, this category might appear to overlap with issues in the Data types and conversion (CW_05) category, but note that they represent different problems. CW_05 contains known issues where Database Migration Service can't convert Oracle code to its PostgreSQL equivalent.

Review date format mask

Potential root cause

Most Oracle date and timestamp format models have appropriate equivalents in PostgreSQL, so the converted code has no semantic or functional differences. Some models don't have an exact match, and their behavior varies. One example is the Oracle RR format that is converted to the PostgreSQL YY format.

Possible mitigation

Review and validate expressions with format model conversions to make sure that the converted code behaves as expected.

Review numeric format mask

Potential root cause

Most source numeric format models have an equivalent in PostgreSQL, and the converted code therefore has no semantic or functional differences. However, some format models might have no exact match or behave slightly differently.

Possible mitigation

Review and validate expressions with format model conversions to make sure that the converted code works as expected.

Review exception code

Potential root cause

When you use RAISE_APPLICATION_ERROR with an error code in the range of -20000 to -20999, Database Migration Service converts it to PostgreSQL RAISE EXCEPTION with a SQLSTATE in the range of CW0000 to CW999. The conversion retains the last 3 digits of the source error code and prefixes it with CW.

Possible mitigation

Review this behavior to determine whether it's suitable for your needs. This review is necessary only if the source error codes are meaningful to your application, support teams, or documentation. If the error code value itself isn't meaningful, you can ignore this warning.

Review exception message

Potential root cause

The SQLERRM function exists both in Oracle PL/SQL and in PostgreSQL, but it might return different error text in each engine. For example, dividing by zero in Oracle returns error text ORA-01476: divisor is equal to zero but ERROR: division by zero in PostgreSQL.

Possible mitigation

If your application, support infrastructure, or documentation depends on the error text, review the conversion. Otherwise, you can ignore this difference.

Review Oracle built-in function emulation

Potential root cause

Database Migration Service code and schema conversion aims to provide Oracle function behavior with PostgreSQL equivalents, but the results might not always be satisfactory for your scenario. As such, conversion workspaces always provide advisory warning with function conversions that might require your review.

Possible mitigation

We recommend you review objects where conversion workspaces emit warnings in the CW_OP0605 issue group.

Review foreign key column data type

Potential root cause

Database Migration Service identified mismatched data type specifications in between parent and child objects (for example, when a parent column is NUMBER(4) and the child column is NUMBER(10)).

Possible mitigation

Most of the time, slight data type mismatches don't cause problems in database functionality. However, we recommend that you review the converted data model for inconsistencies.

Functional review recommended
Potential root cause

This group captures all generic issues that relate to potential functional differences in Oracle and PostgreSQL code. Issues in this group don't fall under any other, more specific, issue groups.

Possible mitigation

Review Oracle built-in function emulation

Potential root cause

Many Oracle built-in functions don't have a direct equivalent in PostgreSQL. To help mitigate that problem for migrations, Database Migration Service converts your code by using different SQL expressions to produce equivalent functional behavior in PostgreSQL.

In some cases, the converted expressions might be complex. Database Migration Service emits warnings in the CW_OP0702 group to highlight possible problems and advise that a function was emulated with an expression.

Possible mitigation

Review the converted code to make sure that the converted functions behave as expected in your PostgreSQL environment.

Autonomous transactions refactoring required

Potential root cause

PostgreSQL doesn't support autonomous transactions.

Possible mitigation

You can achieve autonomous transactions in PostgreSQL by using the dblink, pg_background, or PL/Proxy extension. Database calls made with any one of these extensions execute in a different session, and therefore generate an autonomous transaction.

Database links refactoring required

Potential root cause

Database Migration Service doesn't support Oracle database links. Objects that use links require refactoring.

Possible mitigation

Depending on the target of your database link, you can implement equivalent functionality in PostgreSQL with database extensions, such as dblink, postgres_fdw, oracle_fdw, or PL/Proxy.

Advanced queuing refactoring required

Potential root cause

Oracle Advanced Queuing packages (DBMS_AQ, DBMS_AQADM) don't have PostgreSQL equivalents and require refactoring.

Possible mitigation

Consider the following options:

  • Refactor the queueing functionality away from the database and into the application tier.
  • Use PostgreSQL tables, alerts, and triggers to implement equivalent behavior.
  • Consult your technical solutions team representative for additional assistance.

Database email refactoring required

Potential root cause

Cloud SQL for PostgreSQL doesn't support sending emails directly from the database. Extensions that enable this functionality are also unsupported. As such, Database Migration Service doesn't convert uses of the UTL_SMTP package.

Possible mitigation

Refactor your database email code, and move the responsibility for sending emails to the application tier. You can still use the database to capture the conditions under which sending out email is required.

An example implementation could be to write email details to a dedicated table. This table can also act as an email queue that you poll with a Cloud Run functions function and handle the actual SMTP processing.

Jobs and scheduling refactoring required

Potential root cause

The DBMS_JOB and DBMS_SCHEDULER Oracle packages aren't converted by Database Migration Service. You need to refactor the code that references these packages.

Possible mitigation

For simple jobs without dependencies, you can manually create scheduled jobs in the target PostgreSQL database with the pg_cron extension.

For more complicated schedules that pg_cron doesn't support, you might need to use an application-level or third-party scheduler.

File I/O refactoring required

Potential root cause

Database Migration Service doesn't support the Oracle UTL_FILE package. You need to refactor the code that references this packages.

The Orafce extension includes UTL_FILE emulation, but might not work in a Google managed PostgreSQL environment due to restricted file I/O capabilities.

Possible mitigation
  • Refactor your code to remove dependencies on UTL_FILE.

  • Cloud SQL for PostgreSQL has certain restrictions on file I/O capabilities, so it's not possible to implement this behavior directly in the destination database.

    A possible alternative could be to install PostgreSQL with the orafce extension on a Compute Engine VM in your VPC. You can then use the PL/Proxy extension in your destination database to call the PostgreSQL-compatible UTL_FILE version that runs in the database on the Compute Engine VM.

Synonyms

Potential root cause

PostgreSQL doesn't support synonyms. For code objects, Database Migration Service automatically replaces synonym references with their source schema and object name. If you use synonyms outside of code objects, for example in read-only schemas for database application users, you need to convert them manually.

Possible mitigation

For synonym usage outside of code objects, you can use the PostgreSQL SEARCH_PATH parameter, or refactor your code to use views for query objects.

Global temporary tables

Potential root cause

This issue group is a warning that Database Migration Service detected a global temporary table in your Oracle source code. Migrating global temporary tables requires that you have the pgtt PostgreSQL extension installed and created on the destination database.

Possible mitigation

We recommend that you verify that you have the pgtt PostgreSQL extension installed and created on the destination database.

Review Gemini suggestions
Potential root cause:

This issue group captures all generic errors and warnings related to Gemini-enhanced code conversion.

Possible mitigation: Issues found here might not always point to real problems, but we strongly recommend that you review all Gemini-enhanced conversions to make sure they match your expectations.

Review AI-augmented code
Potential root cause: This DDL code was converted with Gemini-enhanced features and might need your review for accuracy.

Possible mitigation We recommend that you carefully review code converted with AI augmentations to make sure the end result matches the functionality of your source schema.

Citations
Potential root cause: Gemini-enhanced suggestions can include content cited from multiple sources. Certain citations might be subject to license restrictions. We recommend that you review the converted code for citations.

Metadata conversion issues
Potential root cause:

This group captures all conversion issues that don't fall under any other, more specific, issue groups.

Possible mitigation

We recommend that you review the converted code based on your knowledge of the source data model and adjust the code as needed.

Metadata conversion issues
Potential root cause:

This group captures all metadata-tracking issues that don't fall under any other, more specific, issue groups.

Possible mitigation:

Examples of issues in this group are usually related to compilation errors or warnings that can lead to problems with data types in the converted PostgreSQL. We recommend that you review the converted code based on your knowledge of the source data model and adjust the faulty references.

Contact your support team

Potential root cause

In special edge cases, you might encounter an internal error with a valid Oracle source object. If you do, contact your support team for additional assistance.

General conversion issues
Potential root cause

This group contains all issues that don't fall under any other, more specific, issue categories or groups.

Possible mitigation

We recommend that you review the converted code based on your knowledge of the source data model and code, and adjust as needed.