I am attempting to do some batch deletions using Azure's SQL Database, I run a statement as follows:
delete from DbRoles where AccountID = 41;
One would expect that all rows where AccountID is 41, however it only deletes that last matching row from the table. I have found this to be true whether using a regular "where" statement as well as using subqueries like
where Id in (select Id from DbRoles where AccountID = 41)
(And Id in (41, 42, 43, ... etc) also fails) Does anyone have any ideas why this might be?
----- EDIT -----
More info to answer some questions in the comments:
The Roles table does indeed have a trigger on delete (thanks for the tip #Steve Ford) We have a Closure table connected to the Roles table, and it gets updated via triggers when actions occur in the roles table. One of which is the delete trigger, which goes as follows:
CREATE TRIGGER [tr_dbRole_delete]
ON [dbo].[DbRoles]
Instead of DELETE
AS
BEGIN
declare #roleID int;
select #roleID = Id from deleted
--delete self closure
delete from DbRoleClosures
where childID = #roleID and parentID = #roleID
--delete role
delete from DbRoles
where Id = #roleID
END
I think the "Instead" statement might be my culprit, can anyone verify?
Your trigger is only deleting one row because of this statement:
select #roleID = Id from deleted
When you delete multiple rows #roleID will be set to one of the Id's depending upon which order SQL Server queried your deleted table.
Trigger should be this:
CREATE TRIGGER [tr_dbRole_delete]
ON [dbo].[DbRoles]
Instead of DELETE
AS
BEGIN
--delete self closure
delete from DbRoleClosures
where childID = parentId AND
childID in (SELECT Id from deleted)
--delete role
delete from DbRoles
where Id in (SELECT Id from deleted)
END
Related
I have a simple thing to do but somehow can't figure out how to do it.
I have to modify two tables (insert or update) based on existance of a row in the first table.
There is a possibility that some other process will insert the row with id = 1
between getting the flag value and "if" statement that examines its value.
The catch is - I have to change TWO tables based on the flag value.
Question: How can I ensure the atomicity of this operation?
I could lock both tables by "select with TABLOCKX", modify them and release the lock by committing the transaction but ... won't it be overkill?
declare #flag int = 0
begin tran
select #flag = id from table1 where id = 1
if #flag = 0
begin
insert table1(id, ...) values(1, ...)
insert table2(id, ...) values(1, ...)
end
else
begin
update table1 set colX = ... where id = 1
update table2 set colX = ... where id = 1
end
commit tran
To sumarize our conversation and generalize to other's case :
If your column [id] is either PRIMARY KEY or UNIQUE you can put a Lock on that row. No other process will be able to change the value of [id]
If not, in my opinion you won't have other choice than Lock the table with a TABLOCKX. It will prevent any other process to UPDATE,DELETE or INSERT a row.
With that lock, it could possibly allow an other process to SELECT over the table depending on your isolation level.
If your database is in read_committed_snapshot, the other process would read the "old" value of the same [id].
To check your isolation level you can run
SELECT name, is_read_committed_snapshot_on FROM sys.databases
Much more complicated then this, but this is the basic
Person table (id, name, emailaddress)
Salesperson table (id, personID)
CustomerServiceRep table (id, personID)
Jeff is salesperson (id=4) and customerservicerep (id=5) with personID=1.
Simple
Trigger on SalesPerson Table
AFTER DELETE
AS
DECLARE #personID int = (SELECT personID FROM deleted);
IF #personID IS NOT NULL
BEGIN TRY
DELETE FROM Person
WHERE Person.id = #personID;
END TRY
BEGIN CATCH
END CATCH
DELETE FROM SalesPerson WHERE id=4;
Causes
Msg 3616, Level 16, State 1
An error was raised during trigger execution. The batch has been aborted and the user transaction, if any, has been rolled back.
I'm sure there's a much simpler way to not delete personID if it exists from some kind of constraint. Or catch the constraint. To go through every possible table that this could be in seems very repetitive and potentially more difficult when there are more tables/columns that may use this same table/constraint (foreign key).
You need an instead of delete trigger here rather than an after trigger.
CREATE Trigger tr_Delete_person
on Person
INSTEAD OF DELETE
AS
BEGIN
SET NOCOUNT ON;
-- Delete any child records
Delete FROM SalesPerson
WHERE EXISTS (SELECT 1 FROM deleted
WHERE personID = SalesPerson.personID)
Delete FROM CustomerServiceRep
WHERE EXISTS (SELECT 1 FROM deleted
WHERE personID = CustomerServiceRep .personID)
-- Finally delete from the person table
DELETE FROM Person
WHERE EXISTS (SELECT 1 FROM deleted
WHERE personID = Person .personID)
END
You also have a fundamental flaw in your trigger in that you seem to expect that the trigger will be fired once per row - this is NOT the case in SQL Server. Instead, the trigger fires once per statement, and the pseudo table Deleted might contain multiple rows.
Given that that table might contain multiple rows - which one do you expect will be selected here??
DECLARE #personID int = (SELECT personID FROM deleted);
It's undefined - you'll get the value from one, arbitrary row in Deleted, and all others are ignored - typically not what you want!
You need to rewrite your entire trigger with the knowledge the Deleted WILL contain multiple rows! You need to work with set-based operations - don't expect just a single row in Deleted !
Why I can use this select:
SELECT 1 FROM INSERTED
into a trigger, but not run like another select?
I got this error:
Msg 208, Level 16, State 1, Line 1
Invalid object name 'INSERTED'.
It works in this statement (that's inside a trigger):
IF EXISTS(SELECT 1 FROM INSERTED) AND
NOT EXISTS(SELECT 1 FROM DELETED)
BEGIN
-- AFTER INSERT
UPDATE VEIC
SET
VEIC.FLAG = 'I'
FROM
DBVEICULO VEIC
INNER JOIN INSERTED INS ON INS.ID_VEICULO = VEIC.ID_VEICULO
Trigger > http://pastebin.com/9Dh4TUPc
As per the documentation - Create Trigger (Transact-SQL):
DML triggers use the deleted and inserted logical (conceptual) tables. They are structurally similar to the table on which the trigger is defined, that is, the table on which the user action is tried. The deleted and inserted tables hold the old values or new values of the rows that may be changed by the user action. For example, to retrieve all values in the deleted table, use:
So the two tables, INSERTED and DELETED, only exists in the context of a trigger.
I have an SQL Trigger FOR INSERT, UPDATE I created which basically does the following:
Gets a LineID (PrimaryID for the table) and RegionID From the Inserted table and stores this in INT variables.
It then does a check on joining tables to find what the RegionID should be and if the RegionID is not equal what it should be from the Inserted table, then it should update that record.
CREATE TRIGGER [dbo].[TestTrigger]
ON [dbo].[PurchaseOrderLine]
FOR INSERT, UPDATE
AS
-- Find RegionID and PurchaseOrderLineID
DECLARE #RegionID AS INT
DECLARE #PurchaseOrderLineID AS INT
SELECT #RegionID = RegionID, #PurchaseOrderLineID = PurchaseOrderLineID FROM Inserted
-- Find PurchaserRegionID (if any) for the Inserted Line
DECLARE #PurchaserRegionID AS INT
SELECT #PurchaserRegionID = PurchaserRegionID
FROM
(...
) UpdateRegionTable
WHERE UpdateRegionTable.PurchaseOrderLineID = #PurchaseOrderLineID
-- Check to see if the PurchaserRegionID has a value
IF #PurchaserRegionID IS NOT NULL
BEGIN
-- If PurchaserRegionID has a value, compare it with the current RegionID of the Inserted PurchaseOrderLine, and if not equal then update it
IF #PurchaserRegionID <> #RegionID
BEGIN
UPDATE PurchaseOrderLine
SET RegionID = #PurchaserRegionID
WHERE PurchaseOrderLineID = #PurchaseOrderLineID
END
END
The problem I have is that it is not updating the record and I'm guessing, it is because the record hasn't been inserted yet into the PurchaseOrderLine table and I'm doing an update on that. But can you update the row which will be inserted from the Inserted table?
The major problem with your trigger is that it's written in assumption that you always get only one row in INSERTED virtual table.
SQL Server triggers are statement-triggers not row-triggers. You have to take that fact into consideration.
Now if I understand correctly the logic behind this trigger then you need just one update statement in it
CREATE TRIGGER TestTrigger ON PurchaseOrderLine
FOR INSERT, UPDATE
AS
UPDATE l
SET RegionID = u.PurchaserRegionID
FROM PurchaseOrderLine l JOIN INSERTED i
ON l.PurchaseOrderLineID = i.PurchaseOrderLineID JOIN
(
SELECT PurchaseOrderLineID, PurchaserRegionID
FROM UpdateRegionTable -- !!! change this for your proper subquery
) u ON l.PurchaseOrderLineID = u.PurchaseOrderLineID
For this example I've created a fake table UpdateRegionTable. You have to change it to the proper query that returns PurchaseOrderLineID, PurchaserRegionID (in your code you replaced it with ...). Make sure that it returns all necessary rows, not one.
Here is SQLFiddle demo
I think the problem could be that you are making the update to PurchaceOrderLine inside the trigger that is monitoring updates to the same table as well. Try to alter the trigger to just monitor the inserts, than if this works, you can make some changes or break your trigger on two: one for inserts, another for updates.
This has been resolved. I resolved the problem by adding the trigger to another table as the IF #PurchaserRegionID IS NOT NULL was always false.
Question:
In our SQL-Server 2005 database, we have a table T_Groups.
T_Groups has, amongst other things, the fields ID (PK) and Name.
Now some idiot in our company used the name as key in a mapping table...
Which means now one may not alter a group name, because if one does, the mapping is gone...
Now, until this is resolved, I need to add a restriction to T_Groups, so one can't update the group's name.
Note that insert should still be possible, and an update that doesn't change the groupname should also be possible.
Also note that the user of the application & the developers have both dbo and sysadmin rights, so REVOKE/DENY won't work.
How can I do this with a trigger ?
CREATE TRIGGER dbo.yournametrigger ON T_Groups
FOR UPDATE
AS
BEGIN
IF UPDATE(name)
BEGIN
ROLLBACK
RAISERROR('Changes column name not allowed', 16, 1);
END
ELSE
BEGIN
--possible update that doesn't change the groupname
END
END
CREATE TRIGGER tg_name_me
ON tbl_name
INSTEAD OF UPDATE
AS
IF EXISTS (
SELECT *
FROM INSERTED I
JOIN DELETED D ON D.PK = I.PK AND ISNULL(D.name,I.name+'.') <> ISNULL(I.name,D.name+'.')
)
RAISERROR('Changes to the name in table tbl_name are NOT allowed', 16,1);
GO
Depending on your application framework for accessing the database, a cheaper way to check for changes is Alexander's answer. Some frameworks will generate SQL update statements that include all columns even if they have not changed, such as
UPDATE TBL
SET name = 'abc', -- unchanged
col2 = null, -- changed
... etc all columns
The UPDATE() function merely checks whether the column is present in the statement, not whether its value has changed. This particular statement will raise an error using UPDATE() but won't if tested using the more elaborate trigger as shown above.
This is an example of preserving some original values with an update trigger.
It works by setting the values for orig_author and orig_date to the values from the deleted pseudotable each time. It still performs the work and uses cycles.
CREATE TRIGGER [dbo].[tru_my_table] ON [dbo].[be_my_table]
AFTER UPDATE
AS
UPDATE [dbo].[be_my_table]
SET
orig_author = deleted.orig_author
orig_date = deleted.orig_date,
last_mod_by = SUSER_SNAME(),
last_mod_dt = getdate()
from deleted
WHERE deleted.my_table_id IN (SELECT DISTINCT my_table_id FROM Inserted)
ALTER TABLE [dbo].[be_my_table] ENABLE TRIGGER [tru_my_table]
GO
This example will lock any updates on SABENTIS_LOCATION.fk_sabentis_location through a trigger, and will output a detailed message indicating what objects are affected
ALTER TRIGGER dbo.SABENTIS_LOCATION_update_fk_sabentis_location ON SABENTIS_LOCATION
FOR UPDATE
AS
DECLARE #affected nvarchar(max)
SELECT #affected=STRING_AGG(convert(nvarchar(50), a.id), ', ')
FROM inserted a
JOIN deleted b ON a.id = b.id
WHERE a.fk_sabentis_location != b.fk_sabentis_location
IF #affected != ''
BEGIN
ROLLBACK TRAN
DECLARE #message nvarchar(max) = CONCAT('Update values on column fk_sabentis_location locked by custom trigger. Could not update entities: ', #affected);
RAISERROR(#message, 16, 1)
END
Some examples seem to be using:
IF UPDATE(name)
But this seems to evaluate to TRUE if the field is part of the update statement, even if the value itself has NOT CHANGED leading to false positives.