这发生在我身上，因为我做了一个合并，它更新了我正在进行合并的ID。

失败的例子(注意ClownID):

    MERGE ArchitectMain.dbo.BackendClowns AS TGT 
   USING (
    SELECT ClownID
        ,ClownName
        ,Description
        ,Active
        ,EmailSubject
        ,AddedBy
        ,Added
    FROM #temptable1
    ) AS SRC(ClownID, ClownName, Description, Active, EmailSubject, AddedBy, Added)
    ON (TGT.ClownID = SRC.ClownID)
WHEN MATCHED
    THEN
        UPDATE
        SET ClownID = SRC.ClownID
            ,ClownName = SRC.ClownName
            ,Description = SRC.Description
            ,Active = SRC.Active
            ,EmailSubject = SRC.EmailSubject
            ,AddedBy = SRC.AddedBy
            ,Added = SRC.Added;

示例(注意ClownID):

MERGE ArchitectMain.dbo.BackendClowns AS TGT
USING (
    SELECT ClownID
        ,ClownName
        ,Description
        ,Active
        ,EmailSubject
        ,AddedBy
        ,Added
    FROM #temptable1
    ) AS SRC(ClownID, ClownName, Description, Active, EmailSubject, AddedBy, Added)
    ON (TGT.ClownID = SRC.ClownID)
WHEN MATCHED
    THEN
        UPDATE
        SET ClownName = SRC.ClownName
            ,Description = SRC.Description
            ,Active = SRC.Active
            ,EmailSubject = SRC.EmailSubject
            ,AddedBy = SRC.AddedBy
            ,Added = SRC.Added;
            

我做了以下事情:

将相关数据移动到临时存储中 更新主键/标识列值(删除和创建约束) 用新的外键值重新插入相关数据

我把我的解决方案包装在一个存储过程中:

CREATE PROCEDURE [dbo].[UpdateCustomerLocationId]
    @oldCustomerLocationId INT,
    @newCustomerLocationId INT
AS
  /* 
    Updates CustomerLocation.CustomerLocationId @oldCustomerLocationId to @newCustomerLocationId
    
    Example:
        EXEC [dbo].[UpdateCustomerLocationId] 
            @oldCustomerLocationId = 6154874, 
            @newCustomerLocationId = 50334;
    */

BEGIN

    SET NOCOUNT ON;
    
    -- exit if @oldCustomerLocationId does not exists
    IF NOT EXISTS (SELECT * FROM dbo.CustomerLocation cl WHERE cl.CustomerLocationId = @oldCustomerLocationId)
    BEGIN
        PRINT CONCAT('CustomerLocationId ''', @oldCustomerLocationId, ''' (@oldCustomerLocationId) does not exist in dbo.CustomerLocation');
        RETURN 1; -- 0 = success, > 0 = failure
    END

    -- exit if @newCustomerLocationId already exists
    IF EXISTS (SELECT * FROM dbo.CustomerLocation cl WHERE cl.CustomerLocationId = @newCustomerLocationId)
    BEGIN
        PRINT CONCAT('CustomerLocationId ''', @newCustomerLocationId, ''' (@newCustomerLocationId) already exists in dbo.CustomerLocation');
        RETURN 2; -- 0 = success, > 0 = failure
    END
    

    BEGIN TRAN;


        BEGIN -- MOVE related data into temporary storage

            IF EXISTS (SELECT * FROM dbo.CustomerLocationData t WHERE t.CustomerLocationId = @oldCustomerLocationId) BEGIN
    
                IF OBJECT_ID('tempdb..#CustomerLocationData') IS NOT NULL
                    DROP TABLE #CustomerLocationData;

                SELECT * INTO #CustomerLocationData FROM dbo.CustomerLocationData t WHERE t.CustomerLocationId = @oldCustomerLocationId;
                DELETE t FROM dbo.CustomerLocationData t WHERE t.CustomerLocationId = @oldCustomerLocationId;

            END

        END


        BEGIN -- UPDATE dbo.CustomerLocation

            -- DROP CONSTRAINTs
            ALTER TABLE [dbo].[CustomerLocation] DROP CONSTRAINT [UC_CustomerLocation];

            -- INSERT OLD record with new CustomerLocationId
            SET IDENTITY_INSERT dbo.CustomerLocation ON;

            INSERT INTO dbo.CustomerLocation
            (
                CustomerLocationId, CustomerId, LocationId, CustomerLocationIdent, CustomerLocationIdent2, LocationIdent, LocationName, CustomerDistrictId, 
                CustomerLocationGUID, UpdatedOn, IssueManager, EnrollSelfMonitoring, TemperatureControlDeadlineHour, CreatedOn, OperationBegin, ActiveCustomer, 
                Comments, LocationName2, ParentGroup, TempString1, TempString2, TempString3, TempString4, TempString5, AutoInheritFromLocation, ClassificationPrimary
            )
            SELECT @newCustomerLocationId AS CustomerLocationId, CustomerId,LocationId, CustomerLocationIdent, CustomerLocationIdent2, LocationIdent, LocationName, CustomerDistrictId,
                CustomerLocationGUID, UpdatedOn, IssueManager, EnrollSelfMonitoring, TemperatureControlDeadlineHour, CreatedOn, OperationBegin, ActiveCustomer,
                Comments,LocationName2, ParentGroup, TempString1, TempString2, TempString3, TempString4, TempString5, AutoInheritFromLocation, ClassificationPrimary
            FROM dbo.CustomerLocation 
            WHERE CustomerLocationId = @oldCustomerLocationId;

            SET IDENTITY_INSERT dbo.CustomerLocation OFF;

            -- DELETE OLD record
            DELETE cl FROM dbo.CustomerLocation cl WHERE cl.CustomerLocationId = @oldCustomerLocationId;

            -- ADD CONSTRAINTS
            ALTER TABLE [dbo].[CustomerLocation] ADD CONSTRAINT [UC_CustomerLocation] UNIQUE NONCLUSTERED ([CustomerId], [LocationId]);

        END


        BEGIN -- re-INSERT related data from temporary storage

            IF OBJECT_ID('tempdb..#CustomerLocationData') IS NOT NULL BEGIN
                SET IDENTITY_INSERT dbo.CustomerLocationData ON;
                INSERT INTO dbo.CustomerLocationData (Guid, CustomerLocationId, CustomerLocationDataTypeId, Date, Category, Data)
                SELECT Guid, @newCustomerLocationId CustomerLocationId, CustomerLocationDataTypeId, Date, Category, Data FROM #CustomerLocationData;
                SET IDENTITY_INSERT dbo.CustomerLocationData OFF;
            END

        END


    COMMIT TRAN;

END

您可以使用以下代码创建一个新表。

SELECT IDENTITY (int, 1, 1) AS id, column1, column2
INTO dbo.NewTable
FROM dbo.OldTable

然后删除旧的db，并将新db重命名为旧db的名称。注意:columnn1和column2表示旧表中希望保留在新表中的所有列。

如果您特别需要将主键值更改为不同的数字(例如123 -> 1123)。identity属性阻止更改PK值。Set Identity_insert将不起作用。如果您有级联删除，则不建议执行插入/删除操作(除非您关闭了引用完整性检查)。

编辑:新版本的SQL不允许更改syscolumns实体，因此我的部分解决方案必须以艰难的方式完成。关于如何从主键中删除Identity，请参考这个SO: 从表中的列中删除Identity 这个脚本将在PK中关闭标识:

***********************

sp_configure 'allow update', 1
go
reconfigure with override
go


update syscolumns set colstat = 0 --turn off bit 1 which indicates identity column
where id = object_id('table_name') and name = 'column_name'
go


exec sp_configure 'allow update', 0
go
reconfigure with override
go

***********************

接下来，您可以设置关系，以便它们更新外键引用。否则你需要关闭关系强制执行。这个SO链接展示了如何: 如何使用T-SQL临时禁用外键约束?

现在，您可以进行更新。我写了一个简短的脚本来基于相同的列名编写所有的更新SQL(在我的情况下，我需要将CaseID增加1,000,000:

select 
'update ['+c.table_name+'] SET ['+Column_Name+']=['+Column_Name+']+1000000'
from Information_Schema.Columns as c
JOIN Information_Schema.Tables as t ON t.table_Name=c.table_name and t.Table_Schema=c.table_schema and t.table_type='BASE TABLE'
where Column_Name like 'CaseID' order by Ordinal_position

最后，重新启用引用完整性，然后重新启用主键上的Identity列。

注意:我看到有些人在这些问题上问为什么。在我的例子中，我必须将来自第二个生产实例的数据合并到主DB中，这样才能关闭第二个实例。我只需要所有操作数据的PK/ fk不发生冲突。元数据fk是相同的。

我已经解决了这个问题，首先使用DBCC，然后使用插入。例如，如果你的桌子是

首先在表中设置新的当前ID Value为NEW_RESEED_VALUE

MyTable { IDCol, 可乐, colB ｝

    DBCC CHECKIDENT('MyTable', RESEED, NEW_RESEED_VALUE)

然后你可以使用

    insert into MyTable (colA, ColB) select colA, colB from MyTable

这将复制所有记录，但使用新的IDCol值作为NEW_RESEED_VALUE开始。然后，在删除/移动外键引用(如果有)后，可以删除ID值较高的重复行。

就我所见，有几种方法可以做到这一点。，但在我看来最好和最快的方法是以下一种:

标识列有一个计数器，它不一定与已注册的列相同，您可以使用以下SQL命令查看该计数器的值:

DBCC CHECKIDENT('tableName', NORESEED);

然后，如果你想编辑标识列，你将不能，但我建议在重新播种计数器到你需要的数字后，做一个新的寄存器。要重新播种计数器，使用以下命令:

DBCC CHECKIDENT('tableName', RESEED, desiredNumber);