Question

I have three tables: A and B
The relationship is A can have many B
So B has a reference to the A.id as one of its columns

Table A
|id|date|...

Table B
|id|A_id|...

I have created an Oracle trigger on table A so that when it is updated it updates an A_Mod table.

This trigger is

CREATE OR REPLACE TRIGGER TR_A_INSERT_UPDATE
AFTER  INSERT OR UPDATE ON A
FOR EACH ROW
BEGIN
    INSERT INTO A_Mod values(..., :new.date, ...)
END;

This works fine 🙂

My problem si creating the trigger for table B.

The trigger is:

CREATE OR REPLACE TRIGGER TR_B_INSERT_UPDATE
     AFTER  INSERT OR UPDATE ON B
     FOR EACH ROW
     DECLARE
       ts TIMESTAMP;

     BEGIN
   SELECT aa.date INTO ts FROM B bb 
       INNER JOIN A aa ON a.id = bb.A_id
       WHERE bb.id = :new.id;

       INSERT INTO A_Mod values(..., :new.date, ...)
    END;

This trigger is reading the ID of the updated line in table B and then getting the date from the corresponding row in table A. It then tries to insert it into A_Mod

The problem is I get a mutating error

Error report:
SQL Error: ORA-04091: table B is mutating, trigger/function may not see it
ORA-06512: at "TR_B_INSERT_UPDATE", line 5
ORA-04088: error during execution of trigger 'TR_B_INSERT_UPDATE'
04091. 00000 -  "table %s.%s is mutating, trigger/function may not see it"
*Cause:    A trigger (or a user defined plsql function that is referenced in
           this statement) attempted to look at (or modify) a table that was
           in the middle of being modified by the statement which fired it.
*Action:   Rewrite the trigger (or function) so it does not read that table.

Looking at the docs I can remove this error by removing the FOR EACH ROW line and having the trigger fire once per statement rather than once per row. Unfortunately I am using an ORM mapper so do not control how the updates happen. I think there would be times when the update could cover multiple rows.

The docs say something about creating a temporary table but I am not sure how this would help. Would I have to create a temporary table inside the trigger, create a trigger on this temporary table which then updates A_Mod, update this temporary table when the trigger is fired and then delete everything after?

Any tips greatly appreciated.

Thanks

Answer 1

It doesn’t appear that there is any reason in your trigger on B to query the B table. You should be able to simply query A using the :new.a_id

CREATE OR REPLACE TRIGGER TR_B_INSERT_UPDATE
  AFTER  INSERT OR UPDATE ON B
  FOR EACH ROW
DECLARE
   ts TIMESTAMP;
BEGIN
  SELECT a.date 
    INTO ts 
    FROM A a
   WHERE a.id = :new.a_id;

  INSERT INTO A_Mod values(..., :new.date, ...)
END;

From a data modeling standpoint, however, I would be very concerned if a trigger on a child table needed to query any information from the parent table or to insert data into the same history table that the parent table writes to. That seems likely to indicate a normalization issue.