add an identity column to existing table as the primary key and change order - sql

I have a table that has over 7 million records in it. The table does not have a primary key. I would like to add a new identity column and set this as the primary key. I tried adding the column using SSMS, then I set it as the primary key. I called this new column Id.
This almost worked, however I wanted to change the default order of the table to be based another column, for instance a date time column in descending order. Is this possible? Perhaps I need to use a temp table and the ROW_NUMBER() function.
However, I am not very good at SQL. Can someone help?
I also need to have a rollback script so I can get back to the original table.

Here is one more idea:
Step1 - Create temporary clustered index on "date time column in descending order"
CREATE CLUSTERED INDEX ix_YourTableTEMP ON YourTable (DateTimeColumn DESC)
Step2 - Add identity column. Now the IDs should be in order of previously created index - although I don't think there is 100% guarantee on this.
ALTER TABLE YourTable ADD IdColumn INT IDENTITY(1,1)
Step3 - Drop temporary index
DROP INDEX ix_YourTableTEMP ON YourTable
Step4 - Create new clustered PK on new column
ALTER TABLE YourTable
ADD CONSTRAINT PK_YourTable PRIMARY KEY CLUSTERED (IdColumn)

The Link Martin Smith has given is probably the best answer, but here is an alternative:
-- CREATE TABLE WITH SOME DATA IN
CREATE TABLE T (X INT);
INSERT T VALUES (1), (2), (3);
-- CREATE A CLONE OF THIS TABLE, ADDING AN IDENTITY COLUMN
-- USING ORDER BY TO AFFECT THE ORDER OF THE INSERT
SELECT ID = IDENTITY(INT, 1, 1),
T.*
INTO T_Clone
FROM T
ORDER BY X DESC;
-- DROP ORIGINAL TABLE
DROP TABLE T;
-- RENAME CLONE TABLE TO ORIGINAL TABLE NAME
EXECUTE SP_RENAME 'dbo.T_Clone', 'T', 'OBJECT';
-- SELECT FROM TABLE TO CHECK RESULTS
SELECT *
FROM T;
To rollback:
ALTER TABLE T DROP COLUMN ID;
EDIT
It has been pointed out that SELECT ID = IDENTITY(INT, 1, 1).. INTO.. FROM .. ORDER BY ... does not guarantee the order of the insert. So it would appear that the fail safe option is to create your clone table using CREATE TABLE syntax and adding the IDENTITY column:
CREATE TABLE T_Clone
( ID INT IDENTITY(1, 1) NOT NULL PRIMARY KEY,
<your columns>
);
INSERT T_Clone (<your columns>)
SELECT <your columns>
FROM T
ORDER BY ...;
Then carry on with the Drop and rename as above. I can find no documentation to say this method is not reliable for ordering the insert, if it proves to still not be reliable you could use:
SET IDENTITY_INSERT T_Clone ON;
INSERT T_Clone (ID, <your columns>)
SELECT ROW_NUMBER() OVER(ORDER BY ...),
<your columns>
FROM T;
SET IDENTITY_INSERT T_Clone OFF;
Then reseed T_CLone after the insert.

Related

Adding a NOT NULL column to a Redshift table

I'd like to add a NOT NULL column to a Redshift table that has records, an IDENTITY field, and that other tables have foreign keys to.
In PostgreSQL, you can add the column as NULL, fill it in, then ALTER it to be NOT NULL.
In Redshift, the best I've found so far is:
ALTER TABLE my_table ADD COLUMN new_column INTEGER;
-- Fill that column
CREATE TABLE my_table2 (
id INTEGER IDENTITY NOT NULL SORTKEY,
(... all the fields ... )
new_column INTEGER NOT NULL,
PRIMARY KEY(id)
) DISTSTYLE all;
UNLOAD ('select * from my_table')
to 's3://blah' credentials '<aws-auth-args>' ;
COPY my_table2
from 's3://blah' credentials '<aws-auth-args>'
EXPLICIT_IDS;
DROP table my_table;
ALTER TABLE my_table2 RENAME TO my_table;
-- For each table that had a foreign key to my_table:
ALTER TABLE another_table ADD FOREIGN KEY(my_table_id) REFERENCES my_table(id)
Is this the best way of achieving this?
You can achieve this w/o having to load to S3.
modify the existing table to create the desired column w/ a default value
update that column in some way (in my case it was copying from another column)
create a new table with the column w/o a default value
insert into the new table (you must list out the columns rather than using (*) since the order may be the same (say if you want the new column in position 2)
drop the old table
rename the table
alter table to give correct owner (if appropriate)
ex:
-- first add the column w/ a default value
alter table my_table_xyz
add visit_id bigint NOT NULL default 0; -- not null but default value
-- now populate the new column with whatever is appropriate (the key in my case)
update my_table_xyz
set visit_id = key;
-- now create the new table with the proper constraints
create table my_table_xzy_new
(
key bigint not null,
visit_id bigint NOT NULL, -- here it is not null and no default value
adt_id bigint not null
);
-- select all from old into new
insert into my_table_xyz_new
select key, visit_id, adt_id
from my_table_xyz;
-- remove the orig table
DROP table my_table_xzy_events;
-- rename the newly created table to the desired table
alter table my_table_xyz_new rename to my_table_xyz;
-- adjust any views, foreign keys or permissions as required

Update identity column sql [duplicate]

I have a MS SQL 2005 database with a table Test with column ID. ID is an identity column.
I have rows in this table and all of them have their corresponding ID auto incremented value.
Now I would like to change every ID in this table like this:
ID = ID + 1
But when I do this I get an error:
Cannot update identity column 'ID'.
I've tried this:
ALTER TABLE Test NOCHECK CONSTRAINT ALL
set identity_insert ID ON
But this does not solve the problem.
I need to have identity set to this column, but I need to change values as well from time to time. So my question is how to accomplish this task.
You need to
set identity_insert YourTable ON
Then delete your row and reinsert it with different identity.
Once you have done the insert don't forget to turn identity_insert off
set identity_insert YourTable OFF
IDENTITY column values are immutable.
However it is possible to switch the table metadata to remove the IDENTITY property, do the update, then switch back.
Assuming the following structure
CREATE TABLE Test
(
ID INT IDENTITY(1,1) PRIMARY KEY,
X VARCHAR(10)
)
INSERT INTO Test
OUTPUT INSERTED.*
SELECT 'Foo' UNION ALL
SELECT 'Bar' UNION ALL
SELECT 'Baz'
Then you can do
/*Define table with same structure but no IDENTITY*/
CREATE TABLE Temp
(
ID INT PRIMARY KEY,
X VARCHAR(10)
)
/*Switch table metadata to new structure*/
ALTER TABLE Test SWITCH TO Temp;
/*Do the update*/
UPDATE Temp SET ID = ID + 1;
/*Switch table metadata back*/
ALTER TABLE Temp SWITCH TO Test;
/*ID values have been updated*/
SELECT *
FROM Test
/*Safety check in case error in preceding step*/
IF NOT EXISTS(SELECT * FROM Temp)
DROP TABLE Temp /*Drop obsolete table*/
In SQL Server 2012 it is possible to have an auto incrementing column that can also be updated more straightforwardly with SEQUENCES
CREATE SEQUENCE Seq
AS INT
START WITH 1
INCREMENT BY 1
CREATE TABLE Test2
(
ID INT DEFAULT NEXT VALUE FOR Seq NOT NULL PRIMARY KEY,
X VARCHAR(10)
)
INSERT INTO Test2(X)
SELECT 'Foo' UNION ALL
SELECT 'Bar' UNION ALL
SELECT 'Baz'
UPDATE Test2 SET ID+=1
Through the UI in SQL Server 2005 manager, change the column remove the autonumber (identity) property of the column (select the table by right clicking on it and choose "Design").
Then run your query:
UPDATE table SET Id = Id + 1
Then go and add the autonumber property back to the column.
Firstly the setting of IDENTITY_INSERT on or off for that matter will not work for what you require (it is used for inserting new values, such as plugging gaps).
Doing the operation through the GUI just creates a temporary table, copies all the data across to a new table without an identity field, and renames the table.
This can be done using a temporary table.
The idea
disable constraints (in case your id is referenced by a foreign key)
create a temp table with the new id
delete the table content
copy back data from the copied table to your original table
enable previsously disabled constraints
SQL Queries
Let's say your test table have two additional columns (column2 and column3) and that there are 2 tables having foreign keys referencing test called foreign_table1 and foreign_table2 (because real life issues are never simple).
alter table test nocheck constraint all;
alter table foreign_table1 nocheck constraint all;
alter table foreign_table2 nocheck constraint all;
set identity_insert test on;
select id + 1 as id, column2, column3 into test_copy from test v;
delete from test;
insert into test(id, column2, column3)
select id, column2, column3 from test_copy
alter table test check constraint all;
alter table foreign_table1 check constraint all;
alter table foreign_table2 check constraint all;
set identity_insert test off;
drop table test_copy;
That's it.
DBCC CHECKIDENT ( ‘databasename.dbo.orders’,RESEED, 999)
you can change any identity column number with this command,and also you can start that field number from every number you want.for example in my command i ask to start from 1000 (999+1)
hope that it would be enough...good luck
If the column is not a PK you could always create a NEW column in the table with the incremented numbers, drop the original and then alter the new one to be the old.
curious as to why you might need to do this... most I've ever had to futz with Identity columns was to backfill numbers and I just ended up using DBCC CHECKIDENT ( tablename,RESEED,newnextnumber)
good luck!
Identity modifying may fail depending on a number of factors, mainly revolving around the objects/relationships linked to the id column. It seems like db design is as issue here as id's should rarely if ever change (i'm sure you have your reasons and are cascasding the changes). If you really need to change id's from time to time, I'd suggest either creating a new dummy id column that isn't the primary key/autonumber that you can manage yourself and generate from the current values. Alternately, Chrisotphers idea above would be my other suggestion if you're having issues with allowing identity insert.
Good luck
PS it's not failing because the sequential order it's running in is trying to update a value in the list to an item that already exists in the list of ids? clutching at straws, perhaps add the number of rows+1, then if that works subtract the number of rows :-S
If you need to change the IDs occasionally, it's probably best not to use an identity column. In the past we've implemented autonumber fields manually using a 'Counters' table that tracks the next ID for each table. IIRC we did this because identity columns were causing database corruption in SQL2000 but being able to change IDs was occasionally useful for testing.
You can insert new rows with modified values and then delete old rows. Following example change ID to be same as foreign key PersonId
SET IDENTITY_INSERT [PersonApiLogin] ON
INSERT INTO [PersonApiLogin](
[Id]
,[PersonId]
,[ApiId]
,[Hash]
,[Password]
,[SoftwareKey]
,[LoggedIn]
,[LastAccess])
SELECT [PersonId]
,[PersonId]
,[ApiId]
,[Hash]
,[Password]
,[SoftwareKey]
,[LoggedIn]
,[LastAccess]
FROM [db304].[dbo].[PersonApiLogin]
GO
DELETE FROM [PersonApiLogin]
WHERE [PersonId] <> ID
GO
SET IDENTITY_INSERT [PersonApiLogin] OFF
GO
First save all IDs and alter them programmatically to the values you wan't, then remove them from database and then insert them again using something similar:
use [Name.Database]
go
set identity_insert [Test] ON
insert into [dbo].[Test]
([Id])
VALUES
(2)
set identity_insert [Test] OFF
For bulk insert use:
use [Name.Database]
go
set identity_insert [Test] ON
BULK INSERT [Test]
FROM 'C:\Users\Oscar\file.csv'
WITH (FIELDTERMINATOR = ';',
ROWTERMINATOR = '\n',
KEEPIDENTITY)
set identity_insert [Test] OFF
Sample data from file.csv:
2;
3;
4;
5;
6;
If you don't set identity_insert to off you will get the following error:
Cannot insert explicit value for identity column in table 'Test' when
IDENTITY_INSERT is set to OFF.
I saw a good article which helped me out at the last moment .. I was trying to insert few rows in a table which had identity column but did it wrongly and have to delete back. Once I deleted the rows then my identity column got changed . I was trying to find an way to update the column which was inserted but - no luck. So, while searching on google found a link ..
Deleted the columns which was wrongly inserted
Use force insert using identity on/off (explained below)
http://beyondrelational.com/modules/2/blogs/28/posts/10337/sql-server-how-do-i-insert-an-explicit-value-into-an-identity-column-how-do-i-update-the-value-of-an.aspx
Very nice question, first we need to on the IDENTITY_INSERT for the specific table, after that run the insert query (Must specify the column name).
Note: After edit the the identity column, don't forget to off the IDENTITY_INSERT. If you not done, you cannot able to Edit the identity column for any other table.
SET IDENTITY_INSERT Emp_tb_gb_Menu ON
INSERT Emp_tb_gb_Menu(MenuID) VALUES (68)
SET IDENTITY_INSERT Emp_tb_gb_Menu OFF
http://allinworld99.blogspot.com/2016/07/how-to-edit-identity-field-in-sql.html

How we can add column in sql server with auto generated record insertion number. if Index are defined on that table

My database(SQL server 2008) already having some records and having two sorting index on two columns. I tried to add identity column using following query. But it providing me wrong identity numbers associated with records.
alter table Table_name add RECORD_NUMBER int identity(1,1)
record which inserted first having higher number and record which is inserted last having lower number. and some records are having just opposite numbers. Is it possible to remove such kind of problem and insert right insertion order for records.
I did this because i was not having any primary key or unique or compost key in my table. So to identify which record is first entered, i introduce identity column. I was thinking that it will allocate record number according to insertion order.
If it's possible for you to move data in a new table then this is an approach which is basically recommended by Microsoft:
/*************
Preparing for the example, create tables, insert sample data,...
**************/
-- Your existing table
CREATE TABLE dbo.oldtable
(
colA int,
colB nvarchar(10)
)
-- The new table, same structure + new identity column:
CREATE TABLE dbo.newtable
(
colID int IDENTITY(1,1),
colA int,
colB nvarchar(10)
)
--create unique clustered index on new identity column
create unique clustered index ucx_newtable
on dbo.newtable (colID ASC)
--create other indexes as you need them
create index idx_newtable
on dbo.newtable (colA ASC)
--sample data
insert into dbo.oldtable(colA,colB)
values (16,'this'),(17,'is'),(225,'an'),(300,'example')
/*************
Data Movement
**************/
BEGIN TRANSACTION
BEGIN TRY
--move data to new table including new id column which is generated by
--colA, which gives us the order from old to new in this example
SET IDENTITY_INSERT dbo.newtable ON --So we can insert into the colID identity column
INSERT INTO dbo.newtable(colID, colA,colB)
SELECT ROW_NUMBER() OVER(ORDER BY colA ASC) AS colID --this will generate numbers beginning from 1
,colA, colB
FROM dbo.oldtable
SET IDENTITY_INSERT dbo.newtable OFF
--rename old table / or drop it if you want
EXEC sp_rename #objname = 'dbo.oldtable', #newname = 'dbo.oldtable_xyz'
EXEC sp_rename #objname = 'dbo.newtable', #newname = 'dbo.oldtable'
COMMIT TRANSACTION
END TRY
BEGIN CATCH
SELECT ERROR_MESSAGE()
ROLLBACK TRANSACTION
END CATCH

Convert an existing Column to Identity

I have a table in SQL Server with bundle of records. I want to convert the ID column which is Primary Key to an identity Column without loss of data. I thought of the following two approaches:
Create a new table with identity & drop the existing table.
Create a new column with identity & drop the existing column.
but it's clear that they can not be implemented because keeping records is my first priority.
Is there another way to do this?
This solution violates your point 2, but there is no other way and I think your aim is to keep the old values, because nothing else makes sense...
You could do the following:
make it possible to insert into identity columns in your table:
set identity_insert YourTable ON
add a new ID column to your table with identity and insert the values from your old columns
turn identity insert off
set identity_insert YourTable OFF
delete old ID column
rename new column to old name
make it to the primary key
The only problem could be that you have your ID column already connected as foreign key to other tables. Then you have a problem with deleting the old column...
In this case you have to drop the foreign key constraints on your ID column after step 3, then do step 4 to 6 and then recreate your foreign key constraints.
As you are using SQL Server 2012, another possible alternative could be to create a sequence object that has a starting value of the highest ID +1 already in your table, then create a default constraint for your column using GET NEXT VALUE FOR and reference your sequence object you just created.
If you have direct access to the Server Database, just go into the design of the table, select the PK column, and change the identity to "Yes". Make sure you set your seed to the max value of that column. The increment is 1 by default. Save the table design and you should be good to go.
Considering the source table isn't too big:
Create new table (with IDENTITY)
Populate new table from existing table (with IDENTITY_INSERT ON)
Drop old table (drop any existing FKs first)
Rename new table to old name (re-establish FKs if needed)
-- Create Sample Existing Table
DROP TABLE IF EXISTS #tblTest
CREATE TABLE #tblTest
(
ID INT NOT NULL
, Val VARCHAR(10) NOT NULL
)
INSERT INTO #tblTest
(
ID
, Val
)
VALUES
(1, 'a')
, (2, 'b')
, (4, 'c')
GO
-- Create and Populate New Table (with IDENTITY_INSERT ON)
DROP TABLE IF EXISTS #tblTestNew
CREATE TABLE #tblTestNew
(
ID INT IDENTITY(1, 1) NOT NULL
, Val VARCHAR(10) NOT NULL
)
SET IDENTITY_INSERT #tblTestNew ON
INSERT INTO #tblTestNew
(
ID
, Val
)
(
SELECT
#tblTest.ID
, #tblTest.Val
FROM
#tblTest
)
SET IDENTITY_INSERT #tblTestNew OFF
GO
-- Rename Existing Table to Old (can use sp_rename instead, but I can't for temp tables)
SELECT * INTO #tblTestOld FROM #tblTest
DROP TABLE #tblTest
GO
-- Rename New Table to Existing (can use sp_rename instead, but I can't for temp tables)
SELECT * INTO #tblTest FROM #tblTestNew
DROP TABLE #tblTestNew
GO
-- Test Inserting new record
INSERT INTO #tblTest (Val)
VALUES ('d')
-- Verify Results
SELECT * FROM #tblTest
EXEC tempdb.sys.sp_help #objname = N'#tblTest'
-- Drop 'Old' Table (when ready)
DROP TABLE IF EXISTS #tblTestOld
-- Cleanup
DROP TABLE IF EXISTS #tblTest
DROP TABLE IF EXISTS #tblTestNew
DROP TABLE IF EXISTS #tblTestOld
If the table is very large, consider the log growth, Recovery Model, possible single-user mode, etc.
create table t1 (col1 int, col2 varchar(10))
insert into t1 values (10, 'olddata')
--add identity col
alter table t1 add col3 int identity(1,1)
GO
--rename or remove old column
alter table t1 drop column col1
--rename new col to old col name
exec sp_rename 't1.col3', 'col1', 'column'
GO
--add new test , review table
insert into t1 values ( 'newdata')
select * from t1

Inserting a new column for serial number

I have table with 500 records in it and want to insert new column as "serial number" starting with 1.
If you care about the order in which the identity values are assigned, you are best off doing this:
CREATE TABLE dbo.NewTable
(
SerialNumber INT IDENTITY(1,1),
... other columns from original table ...
);
INSERT dbo.NewTable(...other columns...)
SELECT ...other columns...
FROM dbo.OriginalTable
ORDER BY ...ordering criteria...
OPTION (MAXDOP 1); -- to prevent parallelism from messing with identity
DROP TABLE dbo.OriginalTable;
EXEC sp_rename N'dbo.NewTable', N'OriginalTable', N'OBJECT';
You may have to deal with constraints etc. and you will want to do this in a transaction. The point is that just adding an identity column to the table with assign the identity values in an arbitrary order. If you don't care about how the existing values are assigned serial numbers, then just use Kyle's answer.
This could be achieved as follows:
alter table YourTable
add SrNo int identity(1,1)
in PostgreSQL just do:
ALTER TABLE ttaabbllee ADD COLUMN columnName serial NOT NULL; and done!..