Csaba Nagy wrote:
> The reason of the occasional orphan rows is not completely clear to me,
> but it must be some kind of race condition while
> inserting/deleting/?updating concurrently the parent/child tables.
I guess the following sequence would generate a orphaned row.
A: executes "insert into table_child parent_id=1"
B: executes "delete from table_parent where id=1"
A: RI trigger checks for matching row in table_parent
B: The row with id=1 is marked as deleted in table_parent
A: The new row with parent_id=1 is inserted into table_child
B: The delete is commited
A: The insert is comitted.
Any ordering that marks the row as deleted between the execution
of the ri-trigger and the insertion of the new row would lead
to the same result..
greetings, Florian Pflug
In response to
pgsql-hackers by date
|Next:||From: Gavin Sherry||Date: 2007-02-02 12:15:10|
|Subject: Re: Bitmap index thoughts|
|Previous:||From: Csaba Nagy||Date: 2007-02-02 11:01:54|
|Subject: Re: Referential Integrity and SHARE locks|