views:

235

answers:

3

I have a smalldatetime column that I need to alter to be a datetime column. This is something that will be part of an install process, so it cannot be a manual procedure. Unfortunately, the column has a few indexes and a not null constraint on it. The indexes are performance related and would need to be retained only using the new data type. Is it possible to write a statement that will allow me to retain the relevant information while still altering the column datatype? If so, how can this be done?

+1  A: 

EDIT: It depends on the original and changed datatype. If you try to alter a column from varchar to nvarchar, it will fail. Whereas, if you alter column from varchar(16) to varchar(32), it will succeed.

--Disable Index
ALTER INDEX MyIndex ON MyTable DISABLE
GO

-- Change column datatype

--Enable Index
ALTER INDEX MyIndex ON MyTable REBUILD
GO

If you change the type of a column, then all indexes that use that column will have to be rebuilt.

But unless you have huge volumes of data (or run 24/7), rebuilding indexes is no big deal. Just schedule a maintenance window.

Mitch Wheat
Which would be fine, I guess, for Phillip, as long as they do stick around and get rebuilt automatically using the new data type.
Thilo
@Mitch: I did not know about disabling index. That is cool.
Raj More
When I try and run ALTER INDEX [myIndex] on [MyTable] DISABLEI get the error "Incorrect syntax near the keyword 'INDEX'. Am I missing something?
Phillip Benages
Since posting this comment I have found that SQL Server 2000 does not allow you to disable an index. It must be dropped and created. We have a few servers that are still on SQL Server 2000 so the disable command won't work for me.
Phillip Benages
+1  A: 

If you are just changing the size, the Index will still remain on the table.

If you are changing the data type, then you will get an error message stating that objects depend on the column that you are trying to change and therefore you will not be able to change it.

You can script out the indexes in question manually or via script. In SSMS, right click the table and script out the object in question.

If you want programatic index scripting, here is a stored proc that I have been using that I got from an ex colleague of mine.

Drop Proc ScriptIndex
GO
Create Proc ScriptIndex
    @TableName  VarChar (Max),
    @IndexScript VarChar (Max) OUTPUT
AS

-- Get all existing indexes, EXCEPT the primary keys
DECLARE cIX CURSOR FOR
SELECT OBJECT_NAME(SI.Object_ID), SI.Object_ID, SI.Name, SI.Index_ID
FROM Sys.Indexes SI 
    LEFT JOIN INFORMATION_SCHEMA.TABLE_CONSTRAINTS TC 
     ON SI.Name = TC.CONSTRAINT_NAME 
     AND OBJECT_NAME(SI.Object_ID) = TC.TABLE_NAME
WHERE 1=1
    AND OBJECT_NAME(SI.Object_ID) = @TableName
    AND TC.CONSTRAINT_NAME IS NULL
    AND OBJECTPROPERTY(SI.Object_ID, 'IsUserTable') = 1
ORDER BY OBJECT_NAME(SI.Object_ID), SI.Index_ID

DECLARE @IxTable SYSNAME
DECLARE @IxTableID INT
DECLARE @IxName SYSNAME
DECLARE @IxID INT

-- Loop through all indexes
OPEN cIX
FETCH NEXT FROM cIX INTO @IxTable, @IxTableID, @IxName, @IxID
WHILE (@@FETCH_STATUS = 0)
BEGIN
   DECLARE @IXSQL NVARCHAR(4000) 
   DECLARE @PKSQL NVARCHAR(4000) 
   SET @PKSQL = ''
   SET @IXSQL = 'CREATE '

   -- Check if the index is unique
   IF (INDEXPROPERTY(@IxTableID, @IxName, 'IsUnique') = 1)
      SET @IXSQL = @IXSQL + 'UNIQUE '
   -- Check if the index is clustered
   IF (INDEXPROPERTY(@IxTableID, @IxName, 'IsClustered') = 1)
      SET @IXSQL = @IXSQL + 'CLUSTERED '

   SET @IXSQL = @IXSQL + 'INDEX ' + @IxName + ' ON ' + @IxTable + '('

   -- Get all columns of the index
   DECLARE cIxColumn CURSOR FOR 
      SELECT SC.Name
      FROM Sys.Index_Columns IC
         JOIN Sys.Columns SC ON IC.Object_ID = SC.Object_ID AND IC.Column_ID = SC.Column_ID
      WHERE IC.Object_ID = @IxTableID AND Index_ID = @IxID
      ORDER BY IC.Index_Column_ID

   DECLARE @IxColumn SYSNAME
   DECLARE @IxFirstColumn BIT SET @IxFirstColumn = 1

   -- Loop throug all columns of the index and append them to the CREATE statement
   OPEN cIxColumn
   FETCH NEXT FROM cIxColumn INTO @IxColumn
   WHILE (@@FETCH_STATUS = 0)
   BEGIN
      IF (@IxFirstColumn = 1)
         SET @IxFirstColumn = 0
      ELSE
         SET @IXSQL = @IXSQL + ', '

      SET @IXSQL = @IXSQL + @IxColumn

      FETCH NEXT FROM cIxColumn INTO @IxColumn
   END
   CLOSE cIxColumn
   DEALLOCATE cIxColumn

   SET @IXSQL = @IXSQL + ')'
   -- Print out the CREATE statement for the index
   PRINT @IXSQL

   FETCH NEXT FROM cIX INTO @IxTable, @IxTableID, @IxName, @IxID
END

CLOSE cIX
DEALLOCATE cIX

GO
Declare @TableName VarChar (Max), @IndexScript VarChar (Max)

Exec ScriptIndex 'Client', @IndexScript OUTPUT
Print @IndexScript
Raj More
+1  A: 

You can not change the datatype from smalldatetime to datetime with the indexes, unique constraints, foreign key constraints or check constraints in place. You will have to drop them all prior to changing the type. Then:

alter table T alter column TestDate datetime not null

Then recreate the constraints and indexes that still apply.


Some different approaches for generating the drop and creates:

1) If you have given explicit names to all indexes and constraints then your installer can run a static script in each environment (dev, test, user acceptance testing, performance testing, etc, production.)

To generate this explicit script you can: a) Use SSMS (or with SQL Server 2000, enterprise manager) to script the create and drop statements. b) Work from you source code repository to discover the names and definitions of the dependent objects and put together the appropriate static script. c) Attempt to run the alter statement. See what it fails on. Look up the definitions and hand write the drop and create. (Personally, this would be great for writing the drop, not so good at the create.)

2) If you have not given explicit names to all indexes and constraints, then your installer will have to query the data dictionary for the appropriate names and use dynamic SQL to run the drops, in the correct order, prior to the alter column statement and then the creates, in the correct order, after the alter column.

This will be simpler if you know that there are no constraints, and just indexes.

There may be tools or libraries that already know how to do this.

Also, if this is a packaged application, you may not be assured that the local DBAs have not added indexes.

NOTE: If there is a unique constraint, it will have built an index, which you will not be able to drop with DROP INDEX.

Shannon Severance
I was able to find the source sql behind the creation of the indexes and was able to use those to drop those before changing the type. As for the constraint, I found a query through google that can be used to determine the randomly generated constraint name.
Phillip Benages
Declare @constraintName as nvarchar(100)Declare @sql nvarchar(1000) select @constraintName = O.name from sysobjects AS O left join sysobjects AS T on O.parent_obj = T.id where isnull(objectproperty(O.id,'IsMSShipped'),1) = 0 and O.name not like '%dtproper%' and O.name not like 'dt[_]%' and T.name = 'MyTable' and O.name like 'DF__MyTabl__MyCol%' if not @constraintName is null begin select @sql = 'ALTER TABLE [MyTable] DROP CONSTRAINT [' + @constraintName + ']' execute sp_executesql @sql end
Phillip Benages