bad data with Foreign Key constraint

[Lee Wu]

Hi All,



I have found that there are some foreign keys in our detail tables while
there are not primary keys in master tables.

Yes, we have FK constraints on detail tables.

I could not think how it can happen. PG (7.3.2) does not have "disable
constraint" like Oracle, does it?



Sample data from PG:



select rid from c except select rid from r;

rid

-------------

7686396

7688003

7906141

7974816

8033807

(5 rows)



\d c

Table "public.c"

Column | Type |
Modifiers

---------------+--------------------------+-----------------------------
-----------------------

cid | integer | not null default
nextval('c_id_seq'::text)

rid | integer | not null

others...|

Indexes: c_pkey primary key btree (cid),

c_rid_idx btree (rid)

Foreign Key constraints: $1 FOREIGN KEY (rid) REFERENCES r(rid) ON
UPDATE NO ACTION ON DELETE CASCADE



Thank for your help!

[Tom Lane]

Post by Lee Wu
I have found that there are some foreign keys in our detail tables while
there are not primary keys in master tables.
Yes, we have FK constraints on detail tables.
I could not think how it can happen. PG (7.3.2) does not have "disable
constraint" like Oracle, does it?

Hm, you don't have any weird rules on these tables do you? I see
the following in the 7.3.5 commit logs:

2003-10-30 22:57 wieck

* src/: backend/utils/adt/ri_triggers.c,
test/regress/expected/foreign_key.out,
test/regress/sql/foreign_key.sql (REL7_3_STABLE): Fix for possible
referential integrity violation when a qualified ON INSERT rule
split the query into one INSERT and one UPDATE where the UPDATE
then hit's the just created row without modifying the key fields
again. In this special case, the new key slipped in totally
unchecked.

Jan

Also, I find the following in 7.3.3:

2003-03-27 09:33 tgl

* src/: backend/commands/trigger.c, backend/executor/execMain.c,
include/commands/trigger.h (REL7_3_STABLE): GetTupleForTrigger must
use outer transaction's command counter for time qual checking, not
GetCurrentCommandId. Per test case from Steve Wolfe.

I don't recall the implications of this one in detail anymore, but I
think it was triggered by updates inside functions. You might want to
check the archives from back then to see if what Steve was doing looks
anything like stuff your own apps do.

regards, tom lane

---------------------------(end of broadcast)---------------------------
TIP 1: subscribe and unsubscribe commands go to ***@postgresql.org

[Lee Wu]

Hi Tom,

We do not have rules on the tables.

I find manipulating PG dictionary can allow violating RI data to enter
database.

Thanks,

-----Original Message-----
From: pgsql-admin-***@postgresql.org
[mailto:pgsql-admin-***@postgresql.org] On Behalf Of Tom Lane
Sent: Friday, June 18, 2004 4:36 PM
To: Lee Wu
Cc: pgsql-***@postgresql.org
Subject: Re: [ADMIN] bad data with Foreign Key constraint

Post by Lee Wu
I have found that there are some foreign keys in our detail tables

while

Post by Lee Wu
there are not primary keys in master tables.
Yes, we have FK constraints on detail tables.
I could not think how it can happen. PG (7.3.2) does not have "disable
constraint" like Oracle, does it?

Hm, you don't have any weird rules on these tables do you? I see
the following in the 7.3.5 commit logs:

2003-10-30 22:57 wieck

* src/: backend/utils/adt/ri_triggers.c,
test/regress/expected/foreign_key.out,
test/regress/sql/foreign_key.sql (REL7_3_STABLE): Fix for
possible
referential integrity violation when a qualified ON INSERT rule
split the query into one INSERT and one UPDATE where the UPDATE
then hit's the just created row without modifying the key fields
again. In this special case, the new key slipped in totally
unchecked.

Jan

Also, I find the following in 7.3.3:

2003-03-27 09:33 tgl

* src/: backend/commands/trigger.c, backend/executor/execMain.c,
include/commands/trigger.h (REL7_3_STABLE): GetTupleForTrigger
must
use outer transaction's command counter for time qual checking,
not
GetCurrentCommandId. Per test case from Steve Wolfe.

I don't recall the implications of this one in detail anymore, but I
think it was triggered by updates inside functions. You might want to
check the archives from back then to see if what Steve was doing looks
anything like stuff your own apps do.

regards, tom lane

---------------------------(end of broadcast)---------------------------
TIP 1: subscribe and unsubscribe commands go to ***@postgresql.org

---------------------------(end of broadcast)---------------------------
TIP 1: subscribe and unsubscribe commands go to ***@postgresql.org