I am trying to update columns in my child table service after delete on parent table cars using function choose_ideal_car which uses some selects over cars and service tables.
Here is part of my sql script:
CREATE TABLE cars(car_id INT CONSTRAINT pk_id_cars);
CREATE TABLE service(
service_id INT CONSTRAINT pk_id_service PRIMARY KEY,
car_id INT CONSTRAINT fk_id_car REFERENCES cars(car_id) ON DELETE SET NULL,
period VARCHAR2(5) CONSTRAINT check_period CHECK period IN ('even','odd','every'),
period VARCHAR2(3) CONSTRAINT check_day CHECK day IN ('mon','tue','wed','thu','fri')
);
CREATE OR REPLACE TRIGGER service_set_car_id AFTER DELETE ON cars
DECLARE
CURSOR touched_rows is select * from service where car_id = null;
touched_row service%ROWTYPE;
BEGIN
OPEN touched_rows;
LOOP
FETCH touched_rows INTO touched_row;
EXIT WHEN touched_rows%NOTFOUND;
UPDATE service SET car_id = choose_ideal_car(touched_row.period,touched_row.day) WHERE service_id = touched_row.service_id;
END LOOP;
CLOSE touched_rows;
end;
/
For some reason my trigger is never fired.
I also tried creating triggers like this:
CREATE OR REPLACE TRIGGER service_set_car_id AFTER DELETE FROM cars FOR EACH ROW
CURSOR touched_rows is select * from service where car_id = :old.car_id;
...
Which is fired but throws 'mutating error' because function choose_ideal_car uses selects from both tables. Maybe the solution of this is to create duplicate of my cars table and select from it in my choose_ideal_car function instead of selecting from the cars table on which is my trigger defined, but that does not sounds good to me.
While I am writing this post I realized that even if my first trigger is fired it would not work correctly and throw the same 'mutating error'.
So in the end I have two questions:
1) Why is the first trigger never fired?
2) How to fix this mutating error and get all working correctly?
where car_id = null returns no rows, use where car_id is null instead
Related
I have a trigger function:
CREATE OR REPLACE FUNCTION Day_21_bankTriggerFunction()
RETURNS TRIGGER
LANGUAGE plpgsql
AS
$$
DECLARE
act VARCHAR(30);
BEGIN
SELECT account_number INTO act
DELETE FROM depositor
WHERE depositor.account_number = act;
RETURN act;
END;
$$;
\`
and then I have a trigger:
CREATE TRIGGER Day_21_bankTrigger AFTER DELETE on account FOR EACH ROW EXECUTE PROCEDURE Day_21_bankTriggerFunction()
The thought behind this is that is an account was deleted from the account table then this should trigger the function to run and then delete all records on the depositor table where that account is present.
I can create the trigger function and trigger without an issues but if I attempt to delete an account from the account table...I still see the account # in the depositor table when I shouldn't.
Any thoughts?
above is what I tried. Expected results would be to delete an account from the account table and then the trigger function should kick off and remove that account from the depositor table
Usually, relationships of any mode between tables are created by foreign keys. This is the best way and are standards for DBs. Using foreign keys you can control your data. SQL sample:
CREATE TABLE contacts(
contact_id INT GENERATED ALWAYS AS IDENTITY,
customer_id INT,
contact_name VARCHAR(255) NOT NULL,
email VARCHAR(100),
PRIMARY KEY(contact_id),
CONSTRAINT fk_customer
FOREIGN KEY(customer_id)
REFERENCES customers(customer_id)
ON DELETE SET NULL
ON UPDATE SET NULL
);
Third, specify the parent table and parent key columns referenced by the foreign key columns in the REFERENCES clause.
Finally, specify the delete and update actions in the ON DELETE and ON UPDATE clauses.
The delete and update actions determine the behaviors when the primary key in the parent table is deleted and updated. Since the primary key is rarely updated, the ON UPDATE action is not often used in practice. We’ll focus on the ON DELETE action.
PostgreSQL supports the following actions after updating or deleting:
SET NULL (Set values to NULL if data exists on referencing table)
SET DEFAULT (Set values to DEFAULT VALUES of this field if data exists on referencing table)
RESTRICT (Similar to NO ACTION)
NO ACTION (Can not update or delete data if exists on referencing table)
CASCADE (Delete all data if exists on referencing table)
I wrote for you a sample trigger function:
CREATE OR REPLACE FUNCTION Day_21_bankTriggerFunction()
RETURNS trigger
LANGUAGE plpgsql
AS $function$
declare
act varchar(30);
begin
-- get account_number from deleted record
act = old.account_number;
-- SECTION-1 :: Protect deleting if existing data
if (exists(select 1 from depositor where account_number = act)) then
return null;
end if;
-- SECTION-1 :: END
-- SECTION-2 :: Delete all data in the anothers table if exists */
delete from depositor where account_number = act;
return old;
-- SECTION-2 :: END
end
$function$;
CREATE TRIGGER Day_21_bankTrigger
BEFORE DELETE on account
FOR EACH ROW EXECUTE PROCEDURE Day_21_bankTriggerFunction();
Inside my trigger function, I have written two types of SQL codes. (SECTION-1, SECTION-2). You must choose one of them.
I have 2 simple tables:
CREATE TABLE ORDERS
( ORDER_KEY number(10) NOT NULL,
ORDER_NR varchar2(50) NOT NULL,
LAST_UPDATE DATE,
CONSTRAINT ORDERS_PK PRIMARY KEY (ORDER_KEY)
);
CREATE TABLE ORDER_POSITIONS
( ORDER_POS_KEY number(10) NOT NULL,
ORDER_POS_NR number(10),
ORDER_POS_DESCRIPTION varchar2(50),
ORDER_KEY NUMBER(10) NOT NULL,
CONSTRAINT ORDER_POSITIONS_PK PRIMARY KEY (ORDER_POS_KEY),
CONSTRAINT ORDERS_FK
FOREIGN KEY (ORDER_KEY)
REFERENCES ORDERS(ORDER_KEY)
ON DELETE CASCADE
);
On the table ORDER_POSITIONS I created a trigger which should update the column LAST_UPDATE whenever a position is deleted.
CREATE OR REPLACE TRIGGER TGAUDIT_ORDER_POS
AFTER DELETE
ON ORDER_POSITIONS
FOR EACH ROW
DECLARE
BEGIN
UPDATE ORDERS O SET O.LAST_UPDATE = SYSDATE WHERE O.ORDER_KEY = :OLD.ORDER_KEY;
END;
If I delete a position from the table ORDER_POSITION, everything is working fine (the column LAST_UPDATE is updated).
However, if I want to delete an order, all its positions are gonna be deleted, too (via CASCADE DELETE). At this moment also the trigger on the table ORDER_POSITIONS is being raised and it wants to update the column of the table which is currently being deleted - ORDERS. Obviously I get here the error : ORA-04091 Table ORDERS is mutating.
Is there a way to get it solved?
I solved it finally using a compound trigger:
CREATE OR REPLACE TRIGGER TGAUDIT_ORDER_POS
FOR DELETE ON ORDER_POSITIONS
COMPOUND TRIGGER
TYPE parent_key_type IS TABLE OF ORDERS.ORDER_KEY%TYPE;
parent_keys parent_key_type := parent_key_type();
AFTER EACH ROW IS BEGIN
IF DELETING THEN
BEGIN
parent_keys.extend;
parent_keys(parent_keys.last) := :old.ORDER_KEY;
END;
END IF;
END AFTER EACH ROW;
AFTER STATEMENT IS BEGIN
FOR i IN 1..parent_keys.count LOOP
UPDATE DEVART_TEST.ORDERS O SET O.LAST_UPDATE = SYSDATE WHERE O.ORDER_KEY = parent_keys(i);
END LOOP;
END AFTER STATEMENT;
END;
UPDATE: Another option would be to catch this specific exception within the regular trigger.
CREATE OR REPLACE TRIGGER TGAUDIT_ORDER_POS
AFTER DELETE
ON ORDER_POSITIONS
FOR EACH ROW
DECLARE
TABLE_MUTATING EXCEPTION;
PRAGMA EXCEPTION_INIT(TABLE_MUTATING, -4091 );
BEGIN
UPDATE ORDERS O SET O.LAST_UPDATE = SYSDATE WHERE O.ORDER_KEY = :OLD.ORDER_KEY;
EXCEPTION
WHEN TABLE_MUTATING THEN
NULL; -- suppress
END;
I would recreate the foreign-key constraint without the ON DELETE CASCADE clause and delete all order positions before deleting an order. That way you avoid the mutating table error by not having two tables mutating at the same time.
I am trying to implement after insert trigger in PLSQL. The goal is to check if there are multiple (>1) rows having specific status for each client. If so I'd like to rise an exception and roll the insertion back.
I am struggling with implementing warning-free query, which causes error during insertion. How could I manage this?
Here is my implemented trigger which I guess needs some changes.
CREATE TRIGGER blatrigger
AFTER INSERT
ON BLATABLE
REFERENCING NEW AS NEW OLD AS OLD
FOR EACH ROW
DECLARE
exception_name EXCEPTION;
PRAGMA EXCEPTION_INIT (exception_name, -20999);
BEGIN
if (select count(*) as counter from BLATABLE where CLIENTID = :NEW.CLIENTID and STATUS='PENDING').counter > 1
THEN
raise exception_name;
END IF;
END;
Here is the table itself:
create table BLATABLE
(
ID NUMBER(19) not null primary key,
CLIENTID NUMBER(10),
CREATED TIMESTAMP(6),
STATUS VARCHAR2(255 char)
);
The goal is to check if there are multiple (>1) rows having specific status for each client. If so I'd like to rise an exception and roll the insertion back.
No need for a trigger. It looks like a simple unique constraint should get the job done here:
create table blatable (
id number(19) not null primary key,
clientid number(10),
created timestamp(6),
status varchar2(255 char),
constraint blaconstraint unique (clientid, status)
);
The unique constraint prevents duplicates on (clientid, status) across the whole table. If a DML operation (insert, update) attempts to generate a duplicate, an error is raised and the operation is rolled back.
If, on the other end, you want to allow only one "PENDING" status per user, then you can use a unique index as follows:
create unique index bla_index
on blatable( (case when status = 'PENDING' then clientid end) );
Use a Statement Level Trigger, rather than a Row Level by removing FOR EACH ROW, and converting to your code as below :
CREATE OR REPLACE TRIGGER blatrigger
AFTER INSERT ON BLATABLE
REFERENCING NEW AS NEW OLD AS OLD
DECLARE
counter INT;
exception_name EXCEPTION;
PRAGMA EXCEPTION_INIT(exception_name, -20999);
BEGIN
SELECT MAX(COUNT(*))
INTO counter
FROM BLATABLE
WHERE STATUS = 'PENDING'
GROUP BY CLIENTID;
IF counter > 1 THEN
RAISE exception_name;
END IF;
END;
/
where
the SELECT statement need to be removed from IF .. THEN conditional
Most probably, the mutating table error would raise for Row Level Trigger case
Demo
So for now, I created 2 tables, Booking and BookingDetails, I want to create a trigger when I key in the details of BookingDetails it will automatically update totaldays inside the Booking table. Below is my code:
BookingDetails:
create table BookingDetail (
BD_ID int primary key not null,
Date_In date,
Date_Out date,
BK_ID int,
Room_ID int,
foreign key(BK_ID) references Booking(BK_ID),
foreign key(Room_ID) references Room(Room_ID)
)
And also Booking
create table Booking (
BK_ID int primary key not null,
BK_Date Date,
BK_TotalDays int,
BK_PayStatus char(6),
Cus_ID int,
Emp_ID int,
foreign key(Cus_ID) references customer(Cus_ID),
foreign key(Emp_ID) references Employee(Emp_ID)
)
With the function and trigger created:
create function countdays(t1 date, t2 date)
returns INT
return (timestampdiff(16, char(timestamp(t2) - timestamp(t1))))
create trigger totaldays
after insert on bookingdetail
referencing new as n
for each row mode db2sql
update booking
set bk_totaldays =
countdays((select date_in from bookingdetail), (select date_out from
bookingdetail))
where booking.bk_id = n.bk_id;
I have no problem executing these syntax, but when I try to input a new record inside Booking Detail to let the trigger triggers in Booking, errors occured, may I ask why? Thanks in advance.
Look at the information provided by the SQL error:
db2 ? SQL0811
SQL0811N The result of a scalar fullselect, SELECT INTO statement, or
VALUES INTO statement is more than one row.
So this part of your trigger expressions returns more than 1 row
set bk_totaldays = countdays((select date_in from bookingdetail),
(select date_out from bookingdetail))
Fix this to return a single row.
#MichaleTiefenbacher is correct about the cause of the error, but he's wrong about what to do about it. Let's take another look at your trigger again:
create trigger totaldays
after insert on bookingdetail
referencing new as n -- wait, what's this?
for each row mode db2sql
update booking
set bk_totaldays =
countdays((select date_in from bookingdetail), (select date_out from bookingdetail))
where booking.bk_id = n.bk_id;
You have a reference to the value you just inserted! When using FOR EACH ROW, the table reference NEW refers to the singular row just inserted. So you don't even need to look at the full table, just use what you just worked with:
create trigger totaldays
after insert on bookingdetail
referencing new as n
for each row mode db2sql
update booking
set bk_totaldays = countdays(n.date_in, n.date_out)
where booking.bk_id = n.bk_id;
(As I mentioned in my comment to your question, you'll likely want to change how you calculate the days, but that's irrelevant for this)
I am creating a trigger that should compare of the values being inserted with one that already exists in the table. The old referencer doesn't work here because I am inserting, but how can I reference something that already exists?
Here is my tables and trigger:
create table events(eid char(2) primary key, cid char(2));
create table activities(mid char(2), eid char(2),
primary key (mid, eid),
constraint activities_fk foreign key (eid) references events(eid));
create or replace trigger check_valid
before insert or update on activities
for each row when (old.mid=new.mid)
declare
v_eid char(2);
v_cid char(2);
n_cid char(2);
begin
select eid into v_eid from activities
where mid=:new.mid;
select cid into v_cid from events
where eid=v_eid;
select cid into n_cid from events
where eid=:new.eid;
if v_cid=n_cid then
raise_application_error(-20000, 'Error');
end if;
end check_valid;
/
show errors;
You can't generally select from the table you're inserting into in a trigger. This is the mutating table problem, or as I often call it, the "damn mutating table problem".
Basically, don't do this. It's a bad idea. What happens if you have two sessions operating on the table at once? The trigger fires and neither session sees what the other has done until the commit, which is after the trigger. Then you've got unexpected data in your database.
Tom Kyte says, "when I hit a mutating table error, I've got a serious fatal flaw
in my logic."