Conversion issues and patterns reference

Invalid source code

Potential root cause

Errors in this group often occur when Database Migration Service encounters unknown syntax, or when the SQL Server source code isn't valid.

Correct the invalid objects in the source SQL Server 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.

Ensure that all objects you depend on in the migrated code are included in the source tree. You can add the referenced objects to the Database Migration Service source tree, or manually adjust the references in the PostgreSQL code based on your knowledge of the source data model.

If your object refers to an object in another database and uses the three-part entity name, refactor your queries to account for the differences in how SQL Server code and PostgreSQL handle three-part names. Consider the following approaches:

• Modify the application code to connect directly to the other database instead of using three-part names.
• Move the relevant objects to the main database, or create synchronisation mechanism between these databases.
• Use Gemini-enhanced conversion assistance. For more information, see Convert SQL Server code and schema with Gemini assistance .

Missing 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 string (hash) column named md5_hash to the target PostgreSQL table.

You can either retain or remove the md5_hash column after the migration.

Source functionality not yet supported

Potential root cause

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

Find similar functionality in PostgreSQL and modify the converted code accordingly. We recommend that you explore Gemini-powered auto-conversion features for expediting such fixes. For more information, see Convert SQL Server code and schema with Gemini assistance .

SQL Server system view not yet supported

Potential root cause

Database Migration Service doesn't support certain SQL Server system views. For example, you might receive this error for system views related to schema information, such as sys.procedures .

Consider creating an extension in PostgreSQL to provide the missing system view.

SQL Server SQL function not yet supported

Potential root cause

Some SQL Server 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, global system functions starting with @@ ). Some functions might not exist at all.

Inspect which SQL Server 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 .

SQL Server object not yet supported

Potential root cause

Some SQL Server built-in objects are unsupported or only partially supported by Database Migration Service for conversion.

Try to adjust your converted code to replicate the unsupported functionality:

• Create a User-Defined Function (UDF) with the same name as the object in the destination database.
• Replace the object 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 .

Source feature not yet supported

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

Identify the specific SQL Server feature that triggers the error.

• Restructure your code to avoid using the unsupported feature, if possible.
• Replace the unsupported feature with an equivalent that is supported in PostgreSQL.

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

T-SQL feature not yet supported

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

Review and examine the unsupported T-SQL feature that raised the error. Adjust the code based on your knowledge of the T-SQL feature and behavior.

You can also explore Gemini-enhanced conversion assistance. For more information, see Convert SQL Server code and schema with Gemini assistance .

Dynamic SQL options not yet supported

Potential root cause

Database Migration Service has partial support for dynamic SQL conversion. SQL Server EXEC and sp_executesql commands are correctly converted to their PostgreSQL equivalents, but the actual dynamic SQL, DML, or DDL strings aren't converted.

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.

JSON not yet supported

Potential root cause

Database Migration Service code conversion doesn't support SQL Server JSON functions such as JSON_VALUE and JSON_QUERY .

You can rewrite the code to avoid the use of JSON functions or create User-Defined Functions (UDFs) to handle JSON data.

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

Locking and transactions issues

Potential root cause

Database Migration Service code conversion doesn't support the conversion of TRANSACTION and LOCKING statements.

• Verify that transactions are handled appropriately in the rewritten code.

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

XML not yet supported

Potential root cause

Database Migration Service doesn't support SQL Server XML features for conversion.

You must manually convert the XML code in your PostgreSQL schema.

MERGE not yet supported

Potential root cause

The SQL MERGE command is not yet supported by Database Migration Service for conversion.

Modify the converted code to achieve functionally equivalent result. For example, you can:

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

PIVOT not yet supported

Potential root cause

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

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 ARRAY s. For example: SELECT ..., UNNEST(ARRAY['<column1-name>', '<column2-name>']) AS column_name, UNNEST(ARRAY[column1, column2]) AS column_value FROM <table>

ALTER statement option not yet supported

Potential root cause

Database Migration Service doesn't support ALTER statements for code conversion. Objects that contain such statements can raise conversion exceptions.

Replace the SQL Server 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 SQL Server code and schema with Gemini assistance .

SQL feature not yet supported

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.

Adjust the converted code to cover your requirements. You can explore Gemini-powered auto-conversion features to adjust the code. For more information, see Convert SQL Server code and schema with Gemini assistance .

Unsupported syntax

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

We recommend that you review the converted code based on your knowledge of the source syntax and behavior, and adjust the converted code. You can also explore Gemini-enhanced conversion assistance. For more information, see Convert SQL Server 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.

We recommend that you review the converted code based on your knowledge of the source syntax and behavior, and adjust the converted code. You can also explore Gemini-enhanced conversion assistance. For more information, see Convert SQL Server code and schema with Gemini assistance .

Unsupported T-SQL syntax

Potential root cause

Your source code uses T-SQL syntax or elements that are unsupported by Database Migration Service.

We recommend that you review the converted code based on your knowledge of the source syntax and behavior, and adjust the converted code. You can also explore Gemini-enhanced conversion assistance. For more information, see Convert SQL Server 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, for example in comparisons between mismatched data types. For more information on supported objects and data types, see About conversion workspaces .

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 SQL Server code and schema with Gemini assistance .

Unsupported exceptions syntax

Potential root cause

Database Migration Service doesn't support some elements of SQL Server exception handling syntax.

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 SQL Server code and schema with Gemini assistance .

Unsupported collation syntax

Potential root cause

In SQL Server, the COLLATE clause specifies both the code page and sorting rules. Collation rules in PostgreSQL use similar syntax, but are mainly configured through locale settings. This functional discrepancy might cause issues in your converted code.

Inspect the COLLATE clause that raised the error, and verify that the sorting behavior is as you expect in PostgreSQL.

If you need to adjust the code, for newer PostgreSQL versions you can try the CREATE COLLATION command. You can also explore Gemini-enhanced conversion assistance. For more information, see Convert SQL Server 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_SP0500 group captures all generic data type conversion problems that don't fall under other issue groups.

In most cases, Database Migration Service emits an ERROR_UNIMPLEMENTED 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 errors when converting date or timestamp expressions to or from strings based on a format model. You might also see these issues if your data is likely to be affected by the differences between how SQL Server FORMAT() , CAST() , and CONVERT() functions work for dates compared to the PostgreSQL TO_CHAR() function .

The format model in SQL Server is determined by the dateformat setting. You can use the DBCC USEROPTIONS command on the source database to check your format model.

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

Numeric format mask issues

Potential root cause

You might encounter issues or warnings with code that converts strings to numbers based on SQL Server format models. You might also see this issue if your data is likely to be affected by the differences between how SQL Server FORMAT() , CAST() , and CONVERT() functions work for dates compared to the PostgreSQL TO_CHAR() function .

For 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 SQL Server code and schema with Gemini assistance .

Casting issues

Potential root cause

You might encounter errors due to unsupported or inaccurate data type casting. 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.

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 .

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 SQL Server code and schema with Gemini assistance .

Issues in this category represent cases where your source code is correctly converted to the closest PostgreSQL equivalent, but the resulting code might behave in unexpected ways. This happens because of the functional differences in how SQL Server 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 SQL Server code to its PostgreSQL equivalent.

Review date format mask

Potential root cause

Most SQL Server date and timestamp format models have appropriate equivalents in PostgreSQL, so the converted code has no semantic or functional differences. Some format models don't have an exact match, and their behavior varies.

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.

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

Review exception message

Potential root cause

If you use the SQL Server @@ERROR function for error handling, keep in mind that the return number might differ from the converted PostgreSQL SQLERRM . RAISERROR might also cause similar issues and should be replaced by RAISE in most cases. ERROR_NUMBER() and ERROR_MESSAGE() return different numbers, even when replaced by SQLSTATE and SQLERRM .

If your application, support infrastructure, or documentation depends on the error text or error number, a review of the conversion is recommended. Otherwise, you can ignore this message.

Functional review recommended

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

We recommend that you review the converted code based on your knowledge of the source database behvaior, and adjust the code as needed. You can also explore Gemini-powered auto-conversion features for expediting your fixes. For more information, see Convert SQL Server code and schema with Gemini assistance .

Database links refactoring required

Potential root cause

Database Migration Service doesn't support the conversion of SQL Server linked servers .

Refactor the converted code to achieve similar functionality in your destination database. One possible approach is to re-create your database links with the dblink extension .

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.

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.

If the synonym object refers to an object in another database and uses the three-part entity name, refactor your queries to account for the differences in how SQL Server code and PostgreSQL handle three-part names. Consider the following approaches:

• Modify the application code to connect directly to the other database instead of using three-part names.
• Move the relevant objects to the main database, or create synchronisation mechanism between these databases.
• Use Gemini-enhanced conversion assistance. For more information, see Convert SQL Server code and schema with Gemini assistance .

General conversion issues

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

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

Metadata conversion issues

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

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 database behavior 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 SQL Server source object. If you do, contact your support team for additional assistance.