I'm using a blue-green deployment strategy with expand contract database pattern. To achieve that on my database deploy schema I've setted the property BlockOnPossibleDataLoss=true because on Expand phase I can modify my database without any break change with the old version.
I had a column that is not necessary anymore so I followed those steps:
I've changed this column to allow null values
Then my new records don't fill this column anymore
I ran a script that setted null for this column to all table records
Now I need to delete this column, but even with all records with NULL value for this column I can't because I got this error:
Rows were detected. The schema update is terminating because data loss
might occur.'
How can I delete this column even using BlockOnPossibleDataLoss=true?
Create the table with the new schema (without the column you wan't to drop) with a temporary name. Something like tmp_YourTable (Not a temporary table)
Insert all data from the source table, to the newly created table
Drop the source table
Rename the new table, to the old table name. EXEC sp_rename 'tmp_YourTable', 'YourTable';
I need to drop all columns in an existing table in SQL Server. It is possible to delete columns by specifying each column name. I want to delete every column without specifying column name. I am looking for something like
ALTER TABLE tblUser DROP COLUMN *;
Is there any known way to do this?
Answering your question literally, no you can't. If you try to remove the last column, SQL Server will throw the following error:
Msg 4923, Level 16, State 1, Line 12
ALTER TABLE DROP COLUMN failed because 'Id' is the only data column in table 'NoColumns'. A table must have at least one data column.
As such, if you actually want to solve your problem, whatever it is, it would be best to voice the initial problem and not the solution you decided to pursue.
Instead remove all the columns, you could drop the table
DROP TABLE TABLENAME
Or You can mention all the column names in Alter query
ALETR TABLE TableName DROP COLUMN Column1, Column2, Column3....ColumnN
I would like to have a column in my DB accessible via two column names temporarily.
Why? The column name was badly chosen, I would like to refactor it. As I want my webapp to remain stable while changing the column name, it would be good to
have a (let's call it) symlink named better_column_name pointing to the column bad_column_name
change the webapplication to use better_column_name
drop the symlink and rename column to better_column_name
"Refactoring Databases" suggests to actually add a second column which is synchronized on commit in order to achieve this. I am just hoping that there might be an easier way with Oracle, with less work and less overhead.
As long as you have code that uses both column names, I don't see a way to get around the fact that you'll have two (real) columns in that table.
I would add the new column with the correct name and then create a trigger that checks which column has been modified and updates the "other" column correspondingly. So whatever is being updated, the value is synch'ed with the other column.
Once all the code that uses the old column has been migrated, remove the trigger and drop the old column.
Edit
The trigger would so do something like this:
CREATE OR REPLACE TRIGGER ...
...
UPDATE OF bad_column_name, better_column_name ON the_table
...
BEGIN
IF UPDATING ('BAD_COLUMN_NAME') THEN
:new.better_column_name = :new.bad_column_name
END IF;
IF UPDATING ('BETTER_COLUMN_NAME') THEN
:new.bad_column_name = :new.better_column_name
END IF;
END;
The order of the IF statements controls which change has a "higher priority" in case someone updated both columns at the same time.
Rename the table:
alter table mytable rename to mytable_old;
Create a view with the original tablename with both bad_column_name and better_column_name that point to the same column (and of course all the other columns):
create or replace view mytable as
select column1
, column2
, ...
, bad_column_name
, bad_column_name better_column_name
from mytable_old
;
Since this view is updatable by default (I assume here that mytable has a primary key), you can insert/update/delete from the view and it doesn't matter if you use bad_column_name or better_column_name.
After the refactoring, drop the view and rename the table and column:
drop view mytable;
alter table mytable_old rename column bad_column_name to better_column_name;
alter table mytable_old rename to mytable;
The best solution to this is only available in Oracle 11g Release 2: Edition-based Redefinition. This really cool feature allows us to maintain different versions of database tables and PL/SQL code, using special triggers and views. Find out more.
Essentially this is Oracle's built-in implementation of #AHorseWithNoName's suggestion.
you can create a view for the table. And port your application to use that view instead of the table.
create table t (bad_name varchar2(10), c2 varchar2(10));
create view vt as select bad_name AS good_name, c2 from t;
insert into vt (good_name, c2) values ('blub', 'blob');
select * from t;
select * from vt;
If you're on 11g you could look at using a virtual column. I'd probably be tempted to change the order slightly; rename the real column and create the virtual one using the old (bad) name, which can then be dropped at leisure. You may be restricted, of course, and there may be implications on other objects being invalidated that make this order less suitable for you.
We have a production table with 770 million rows and change. We want(/need?) to change the Primary ID column from int to bigint to allow for future growth (and to avoid the sudden stop when the 32bit integer space is exhausted)
Experiments in DEV have shown that this is not as simple as altering the column as we would need to drop the index and then re-create it. So far in DEV (which is a bit humbler than PROD) the dropping of the index has not finished after 1 and a half hours. This table is hit 24/7 and having it offline for such a long time is not an option.
Has anyone else had to deal with a similar situation? How did you get it done?
Are there alternatives?
Edit: Additional Info:
The Primary key is clustered.
You could attempt a staged approach.
Create a new bigint column
Create an insert trigger to keep new entries in sync with the 2 columns
Execute an update to populate all the empty values in the bigint column with the converted value
Change the primary index on the table from your old id column to the new one
Point any FK's and queries to use the new column
Change the new column to become your identity column and remove the insert trigger from #2
Delete the old ID column
You should end up spreading the pain out over these 7 steps instead of hitting it all at once.
Create a parallel table with the longer data type for new rows and UNION the results?
What I had to do was copy the data into a new table with the desired structure (primary/clustered key only, non-clustered/FK once complete). If you don't have the room, you could bcp out the data and back in. You may need an application outage to make this happen.
What doesn't work: alter table Orderhistory alter column ID bigint because of the primary key. Don't drop the key and alter column as you will just fill your log file and take much longer than copy/bcp.
Never use the SSMS tools designer to change a column property, it copies table into temp table then does a rename once done. Lookup the alter table alter column syntax and use it and possibly defrag once complete if you modified a column wider that sits in middle of table.
When adding a column to an existing table, Oracle always puts the column at the end of the table. Is it possible to tell Oracle where it should appear in the table? If so, how?
The location of the column in the table should be unimportant (unless there are "page sizes" to consider, or whatever Oracle uses to actually store the data). What is more important to the consumer is how the results are called, i.e. the Select statement.
rename YOUR_ORIGINAL_TABLE as YOUR_NEW_TABLE;
create table YOUR_ORIGINAL_TABLE nologging /* or unrecoverable */
as
select Column1, Column2, NEW_COLUMN, Column3
from YOUR_NEW_TABLE;
Drop table YOUR_NEW_TABLE;
Select * From YOUR_ORIGINAL_TABLE; <<<<< now you will see the new column in the middle of the table.
But why would you want to do it? It's seems illogical. You should never assume column ordering and just use named column list if column order is important.
Why does the order of the columns matter? You can always alter it in your select statement?
There's an advantage to adding new columns at the end of the table. If there's code that naively does a "SELECT *" and then parses the fields in order, you won't be breaking old code by adding new columns at the end. If you add new columns in the middle of the table, then old code may be broken.
At one job, I had a DBA who was super-anal about "Never do 'SELECT *'". He insisted that you always write out the specific fields.
What I normally do is:
Rename the old table.
Create the new table with columns in the right order.
Create the constraints for that new table.
Populate with data:Insert into new_table select * from renamed table.
I don't think that this can be done without saving the data to a temporary table, dropping the table, and recreating it. On the other hand, it really shouldn't matter where the column is. As long as you specify the columns you are retrieving in your select statement, you can order them however you want.
Bear in mind that, under the tables, all the data in the table records are glued together. Adding a column to the end of a table [if it is nullable or (in later versions) not null with a default] just means a change to the table's metadata.
Adding a column in the middle would require re-writing every record in that table to add the appropriate value (or markers) for that column. In some cases, that might mean the records take up more room on the blocks and some records need to be migrated.
In short, it's a VAST amount of IO effort for a table of any real size.
You can always create a view over the table that has the columns in the preferred order and use that view in a DML statement just as you would the table
I don't believe so - SQL Server doesn't allow these either. The method I always have to use is:
Create new table that looks right (including additional column
Begin transaction
select all data from old table into new one
Drop old table
Rename new table
Commit transaction.
Not exactly pretty, but gets the job done.
No, its not possible via an "ALTER TABLE" statement. However, you could create a new table with the same definition as your current one, albeit with a different name, with the columns in the correct order in the way you want them. Copy the data into the new table. Drop the old table. Rename the new table to match the old table name.
Tom Kyte has an article on this on AskTom
link text
Apparently there's a trick involving marking the "after" columns INVISIBLE; when restored, they end up at the back.
CREATE TABLE yourtable (one NUMBER(5, 0), two NUMBER(5, 0), three NUMBER(5, 0), four NUMBER(5, 0))
ALTER TABLE yourtable ADD twopointfive NUMBER(5, 0);
ALTER TABLE yourtable MODIFY (three INVISIBLE, four INVISIBLE);
ALTER TABLE yourtable MODIFY (three VISIBLE, four VISIBLE);
https://oracle-base.com/articles/12c/invisible-columns-12cr1#invisible-columns-and-column-ordering
1) Ok so you can't do it directly. We don't need post after post saying the same thing, do we?
2) Ok so the order of columns in a table doesn't technically matter. But that's not the point, the original question simply asked if you could or couldn't be done. Don't presume that you know everybody else's requirements. Maybe they have a table with 100 columns that is currently being queried using "SELECT * ..." inside some monstrously hacked together query that they would just prefer not to try to untangle, let alone replace "*" with 100 column names. Or maybe they are just anal about the order of things and like to have related fields next to each other when browsing schema with, say SQL Developer. Maybe they are dealing with non-technical staff that won't know to look at the end of a list of 100 columns when, logically, it should be somewhere near the beginning.
Nothing is more irritating than asking an honest question and getting an answer that says: "you shouldn't be doing that". It's MY job, not YOURS! Please don't tell me how to do my job. Just help if you can. Thanks!
Ok... sorry for the rant. Now...at www.orafaq.com it suggests this workaround.
First suppose you have already run:
CREATE TABLE tab1 ( col1 NUMBER );
Now say you want to add a column named "col2", but you want them ordered "col2", "col1" when doing a "SELECT * FROM tbl1;"
The suggestion is to run:
ALTER TABLE tab1 ADD (col2 DATE);
RENAME tab1 TO tab1_old;
CREATE TABLE tab1 AS SELECT 0 AS col1, col1 AS col2 FROM tab1_old;
I found this to be incredibly misleading. First of all, you're filling "col1" with zero's so, if you had any data, then you are losing it by doing this. Secondly, it's actually renaming "col1" to "col2" and fails to mention this. So, here's my example, hopefully it's a little clearer:
Suppose you have a table that was created with the following statement:
CREATE TABLE users (first_name varchar(25), last_name varchar(25));
Now say you want to insert middle_name in between first_name and last_name. Here's one way:
ALTER TABLE users ADD middle_name varchar(25);
RENAME users TO users_tmp;
CREATE TABLE users AS SELECT first_name, middle_name, last_name FROM users_tmp;
/* and for good measure... */
DROP TABLE testusers_tmp;
Note that middle_name will default to NULL (implied by the ALTER TABLE statement). You can alternatively set a different default value in the CREATE TABLE statement like so:
CREATE TABLE users AS SELECT first_name, 'some default value' AS middle_name, last_name FROM users_tmp;
This trick could come in handy if you're adding a date field with a default of sysdate, but you want all of the existing records to have some other (e.g. earlier) date value.