Reputation: 6923
Given a simple table, with an ID what is the correct way to audit a column being changed. I am asking after looking after various answers which seem not to be working.
Here is what I have:
Create Table Tbl_Audit
(
AuditId int identity(1,1) not null,
Tbl_Id int not null.
Tbl_Old_ColumnValue varchar(255),
Tbl_New_ColumnValue varchar(255)
)
GO
Create Trigger Tr_Tbl_ColumnChanged on Tbl
after insert, update
As
begin
if(update(ColumnName))
begin
insert into tbl_audit
(
Tbl_Id,
Tbl_Old_ColumnName,
Tbl_New_ColumnName
)
select
tbl.PKId,
tbl.ColumnName,
i.ColumnName,
from
Tbl tbl join
inserted i
on tbl.PKId = i.PKId
end
What I see is thousands of examples where Tbl_Old_ColumnValue
= Tbl_New_ColumnValue
, which is not what I want.
I would expect to run:
select top 10 * from tbl_audit where Tbl_Old_ColumnValue !=Tbl_New_ColumnValue
But this returns no results.
In order to get results of columns that actually changed, I need to run a very expensive query:
select top 10
old.AuditId,
old.Tbl_Old_ColumnValue,
new.Tbl_Old_ColumnValue as [Tbl_New_ColumnValue]
from tbl_audit [old]
join Tbl_Audit [new]
on [ol].Tbl_Id= [new].Tbl_Id and [old].AuditId != [new].AuditId
where [old].Tbl_Old_ColumnValue != [new].Tbl_Old_ColumnValue
Results:
AuditId Tbl_Id Tbl_Old_ColumnValue Tbl_New_ColumnValue
10051 1 old_value old_value
10052 1 new_value new_value
But that doesn't produce what I expect:
AuditId Tbl_Id Tbl_Old_ColumnValue Tbl_New_ColumnValue
10057 1 old_value Some New Value
Oddly, If I modify the column directly via SSMS using:
update Tbl set Tbl.ColumnValue = 'Some New Value'
I see what I expect from my trigger:
AuditId Tbl_Id Tbl_Old_ColumnValue Tbl_New_ColumnValue
10057 1 old_value Some New Value
What am I doing wrong?
Also, how do I eliminate auditing of row where update(ColumnName)
is actually false. IE, the ColumnName (even if being set) is not audit when it is being set to the previous/old value.
Upvotes: 0
Views: 740
Reputation: 27202
update(ColumnName)
doesn't mean that the value has changed, just that that column was involved in the insert/update - and it will always be involved in an insert. You need to compare the old and new values using inserted
and deleted
e.g.
insert into tbl_audit
(
Tbl_Id,
Tbl_Old_ColumnName,
Tbl_New_ColumnName
)
select
tbl.PKId,
tbl.ColumnName,
i.ColumnName,
from
inserted i
left join deleted d on d.PKId = i.PKId
-- Insert d.PKId is null, there are no records in deleted
where d.PKId is null
-- Change from null to value
or (i.ColumnName is null and d.ColumnName is not null)
-- Change from value to null
or (i.ColumnName is not null and d.ColumnName is null)
-- Change in value
or i.ColumnName <> d.ColumnName;
You can potentially simplify the null check using coalesce
and a suitable value which will never actually occur in your data.
The documentation is actually pretty good on all this.
And if the column is not always included in an update, then the update(ColumnName)
test is still worth doing because it speeds up the trigger, and triggers should be as fast as possible. Personally I short circuit out early e.g. if not update(ColumnName) return;
Obviously you need to adapt that logic to handle all the columns you are auditing.
Upvotes: 3