Since the release of SQL Server 2005, you’ve been able to handle errors in your T-SQL code by including a TRY…CATCH block that controls the flow of your script should an error occur, similar to how procedural languages have traditionally handled errors. The TRY…CATCH block makes it easy to return or audit error-related data, as well as take other actions. And within the block-specifically, the CATCH portion-you’ve been able to include a RAISERROR statement in order to re-throw error-related data to the calling application. However, with the release of SQL Server 2012, you now have a replacement for RAISERROR, the THROW statement, which makes it easier than ever to capture the error-related data.
In this article, we’ll look at the TRY…CATCH block used with both the RAISERROR and THROW statements. The examples are based on a table I created in the AdventureWorks2012 sample database, on a local instance of SQL Server 2012. Listing 1 shows the T-SQL script I used to create the LastYearSales table.
1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 |
USE AdventureWorks2012; GO IF OBJECT_ID('LastYearSales', 'U') IS NOT NULL DROP TABLE LastYearSales; GO SELECT BusinessEntityID AS SalesPersonID, FirstName + ' ' + LastName AS FullName, SalesLastYear INTO LastYearSales FROM Sales.vSalesPerson WHERE SalesLastYear > 0; GO |
The script should be fairly straightforward. I use a SELECT…INTO statement to retrieve data from the Sales.vSalesPerson view and insert it into the newly created table. However, to demonstrate how to handle errors, we need to add one more element to our table: a check constraint that ensures the SalesLastYear value is never less than zero. Listing 2 shows the ALTERTABLE statement I used to add the constraint.
1 2 3 |
ALTER TABLE LastYearSales ADD CONSTRAINT ckSalesTotal CHECK (SalesLastYear >= 0); GO |
The constraint makes it easy to generate an error when updating the table. All I have to do is try to add a negative amount to the SalesLastYear column, an amount large enough to cause SQL Server to throw an error. Once we’ve created our table and added the check constraint, we have the environment we need for the examples in this article. You can just as easily come up with your own table and use in the examples. Just be sure you have a way of violating a constraint or you come up with another mechanism to generate an error. The goal is to create a script that handles any errors.
Working with the TRY…CATCH Block
Once we’ve set up our table, the next step is to create a stored procedure that demonstrates how to handle errors. The procedure, UpdateSales, modifies the value in the SalesLastYear column in the LastYearSales table for a specified salesperson. It works by adding or subtracting an amount from the current value in that column. Listing 3 shows the script I used to create the procedure. Notice that I include two input parameters-@SalesPersonID and @SalesAmt-which coincide with the table’s SalesPersonID and SalesLastYear columns.
1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 |
USE AdventureWorks2012; GO IF OBJECT_ID('UpdateSales', 'P') IS NOT NULL DROP PROCEDURE UpdateSales; GO CREATE PROCEDURE UpdateSales @SalesPersonID INT, @SalesAmt MONEY = 0 AS BEGIN BEGIN TRY BEGIN TRANSACTION; UPDATE LastYearSales SET SalesLastYear = SalesLastYear + @SalesAmt WHERE SalesPersonID = @SalesPersonID; COMMIT TRANSACTION; END TRY BEGIN CATCH IF @@TRANCOUNT > 0 ROLLBACK TRANSACTION; DECLARE @ErrorNumber INT = ERROR_NUMBER(); DECLARE @ErrorLine INT = ERROR_LINE(); DECLARE @ErrorMessage NVARCHAR(4000) = ERROR_MESSAGE(); DECLARE @ErrorSeverity INT = ERROR_SEVERITY(); DECLARE @ErrorState INT = ERROR_STATE(); PRINT 'Actual error number: ' + CAST(@ErrorNumber AS VARCHAR(10)); PRINT 'Actual line number: ' + CAST(@ErrorLine AS VARCHAR(10)); RAISERROR(@ErrorMessage, @ErrorSeverity, @ErrorState); END CATCH END; GO |
The main body of the procedure definition, enclosed in the BEGIN…END block, contains the TRY…CATCH block, which itself is divided into the TRY block and the CATCH block. The TRY block starts with BEGINTRY and ends with ENDTRY and encloses the T-SQL necessary to carry out the procedure’s actions. In this case, I include an UPDATE statement that adds the @SalesAmount value to the SalesLastYear column. The statement is enclosed in BEGINTRANSACTION and COMMITTRANSACTION statements to explicitly start and commit the transaction. Examples vary in terms of where they include the transaction-related statements. (Some don’t include the statements at all.) Just keep in mind that you want to commit or rollback your transactions at the appropriate times, depending on whether an error has been generated.
If the UPDATE statement runs successfully, the SalesLastYear value is updated and the operation is completed, in which case, the code in the CATCH block is never executed. However, if the UPDATE statement fails and SQL Server generates an error, the transaction is terminated and the database engine jumps to the CATCH block. The CATCH block starts with BEGINCATCH and ends with ENDCATCH and encloses the statements necessary to handle the error.
For the stored procedure in Listing 3, the first step I take in the CATCH block is to roll back the transaction if it is still running. I start by using the @@TRANCOUNT function to determine whether any transactions are still open. @@TRANCOUNT is a built-in SQL Server function that returns the number of running transactions in the current session. In this case, there should be only one (if an error occurs), so I roll back that transaction.
Next, I declare a set of variables based on system functions that SQL Server makes available within the scope of the CATCH block. The functions return error-related information that you can reference in your T-SQL statements. Currently, SQL Server supports the following functions for this purpose:
- ERROR_NUMBER(): The number assigned to the error.
- ERROR_LINE(): The line number inside the routine that caused the error.
- ERROR_MESSAGE(): The error message text, which includes the values supplied for any substitutable parameters, such as times or object names.
- ERROR_SEVERITY(): The error’s severity.
- ERROR_STATE(): The error’s state number.
- ERROR_PROCEDURE(): The name of the stored procedure or trigger that generated the error.
For this example, I use all but the last function, though in a production environment, you might want to use that one as well.
After I declare the variables, I include two PRINT statements that display the values of the @ErrorNumber and @ErrorLine variables (along with some explanatory text). The reason I do this is to demonstrate the difference between what the actual values are and what the RAISERROR statement returns, as you’ll see shortly.
The RAISERROR statement comes after the PRINT statements. The statement returns error information to the calling application. Generally, when using RAISERROR, you should include an error message, error severity level, and error state. The rules that govern the RAISERROR arguments and the values they return are a bit complex and beyond the scope of this article, but for the purposes of this example, I simply pass in the @ErrorMessage, @ErrorSeverity, and @ErrorState variables as arguments.
NOTE: For more information about the RAISERROR statement, see the topic “RAISERROR (Transact-SQL)” in SQL Server Books Online.
That’s basically all you need to do to create a stored procedure that contains a TRY…CATCH block. In a moment, we’ll try out our work. But first, let’s retrieve a row from the LastYearSales table to see what the current value is for salesperson 288. Listing 4 shows the SELECT statement I used to retrieve the data.
1 2 3 |
SELECT FullName, SalesLastYear FROM LastYearSales WHERE SalesPersonID = 288 |
Not surprisingly, the statement returns the name and total sales for this salesperson, as shown in Listing 5. As you can see, Rachel Valdez shows over $1.3 million dollars in sales for last year.
1 2 |
FullName SalesLastYear Rachel Valdez 1307949.7917 |
Now let’s try out the UpdateSales stored procedure. Just for fun, let’s add a couple million dollars to Rachel Valdez’s totals. Listing 6 shows how I use the EXEC statement to call the procedure and pass in the salesperson ID and the $2 million.
1 |
EXEC UpdateSales 288, 2000000; |
The stored procedure should run with no problem because we’re not violating the check constraint. If we were to execute the SELECT statement again (the one in Listing 4), our results would look similar to those shown in Listing 7. Notice all the extra cash.
1 2 |
FullName SalesLastYear Rachel Valdez 3307949.7917 |
Now let’s look what happens if we subtract enough from her account to bring her totals to below zero. In listing 8, I run the procedure once again, but this time specify -4000000 for the amount.
1 |
EXEC UpdateSales 288, -4000000; |
As you’ll recall, after I created the LastYearSales table, I added a check constraint to ensure that the amount could not fall below zero. As a result, the stored procedure now generates an error, which is shown in Listing 9.
1 2 3 4 5 |
(0 row(s) affected) Actual error number: 547 Actual line number: 9 Msg 50000, Level 16, State 0, Procedure UpdateSales, Line 27 The UPDATE statement conflicted with the CHECK constraint "ckSalesTotal". The conflict occurred in database "AdventureWorks2012", table "dbo.LastYearSales", column 'SalesLastYear'. |
As expected, the information we included in the CATCH block has been returned. But notice that the actual error number (547) is different from the RAISERROR message number (50000) and that the actual line number (9) is different from the RAISERROR line number (27). In theory, these values should coincide. But as I mentioned earlier, the rules that govern RAISERROR are a bit quirky.
Working with the THROW Statement
To simplify returning errors in a CATCH block, SQL Server 2012 introduced the THROW statement. With the THROW statement, you don’t have to specify any parameters and the results are more accurate. You simply include the statement as is in the CATCH block.
NOTE: You can use the THROW statement outside of the CATCH block, but you must include parameter values to do so. For more information about the THROW statement, see the topic “THROW (Transact-SQL)” in SQL Server Books Online.
To demonstrate the THROW statement, I defined an ALTER PROCEDURE statement that modifies the UpdateSales procedure, specifically the CATCH block, as shown in Listing 10.
1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 |
ALTER PROCEDURE UpdateSales @SalesPersonID INT, @SalesAmt MONEY = 0 AS BEGIN BEGIN TRY BEGIN TRANSACTION; UPDATE LastYearSales SET SalesLastYear = SalesLastYear + @SalesAmt WHERE SalesPersonID = @SalesPersonID; COMMIT TRANSACTION; END TRY BEGIN CATCH IF @@TRANCOUNT > 0 ROLLBACK TRANSACTION; DECLARE @ErrorNumber INT = ERROR_NUMBER(); DECLARE @ErrorLine INT = ERROR_LINE(); PRINT 'Actual error number: ' + CAST(@ErrorNumber AS VARCHAR(10)); PRINT 'Actual line number: ' + CAST(@ErrorLine AS VARCHAR(10)); THROW; END CATCH END; GO |
Notice that I retain the @ErrorNumber and @ErrorLine variable declarations and their related PRINT statements. I do so only to demonstrate the THROW statement’s accuracy. In actually, I need only to roll back the transaction and specify the THROW statement, without any parameters.
Now let’s execute the stored procedure again, once more trying to deduct $4 million from the sales amount, as shown in Listing 11.
1 |
EXEC UpdateSales 288, -4000000; |
Once again, SQL Server returns an error. Only this time, the information is more accurate. As you can see in Listing 12, the message numbers and line numbers now match. No longer do we need to declare variables or call system functions to return error-related information to the calling application.
1 2 3 4 5 |
(0 row(s) affected) Actual error number: 547 Actual line number: 8 Msg 547, Level 16, State 0, Procedure UpdateSales, Line 8 The UPDATE statement conflicted with the CHECK constraint "ckSalesTotal". The conflict occurred in database "AdventureWorks2012", table "dbo.LastYearSales", column 'SalesLastYear'. |
As you can see, SQL Server 2012 makes handling errors easier than ever. Even if you’ve been using the TRY…CATCH block for a while, the THROW statement should prove a big benefit over RAISERROR. And if you’re new to error handling in SQL Server, you’ll find that the TRY…CATCH block and the THROW statement together make the process a fairly painless one, one well worth the time and effort it takes to learn and implement them.
Load comments