Re: BUG #2379: Duplicate pkeys in table

Philip Warner <pjw(at)rhyme(dot)com(dot)au> writes:
> Tom Lane wrote:
>> The thing that is striking though is
>> that the xmin/cmin values are all the same, indicating that all six
>> tuples were inserted by the same command. That seems pretty odd. Can
>> you show us the procedure by which rows are inserted in this table?
>>
> The original insertion is probably not relevant (it happened months
> ago); there are many places that update the table.

OK, what I should say is that all these tuples were updated by the same
command.

> Updates happen regularly from many sources, but the procedure that does
> the most updates is a trigger. Do you want to see that?

Please.

Also, if you care to run pg_filedump -i -F over the table, it'd be
interesting to see the complete header info for each of these tuples.

regards, tom lane