Re: Getting rid of duplicate tables.

Tom Lane wrote:

>Jared Carr <jared(at)89glass(dot)com> writes:
>
>
>>>For starters, what PG version is this? And how many entries are there
>>>in pg_class at the moment? (The output from "VACUUM VERBOSE pg_class"
>>>would be useful to look at.)
>>>
>>>
>
>
>
>>This is version 7.4.1 (upgraded from 7.4 about 3 weeks ago).
>>
>>
>
>Okay, that shoots down my faint hope that this was an old bug. I am
>growing increasingly suspicious that there is some corner-case problem
>with duplicate rows in 7.4, because we've seen two or three reports now
>of problems with unexpected duplicates. The reports all have different
>symptoms but I sense a common theme ...
>
>

Well, hopefully I can provide enough information to find out what that
theme is.

>>And the contents of pg_class relating to this table.
>>
>>live=# select * from pg_class where relname='order_to_do';
>>
>>
>
>I was just about to ask for that, but I need to see the system columns
>too:
>
>select oid,ctid,cmax,xmax,cmin,xmin,* from pg_class
> where relname='order_to_do';
>
>

live=# select oid,ctid,cmax,xmax,cmin,xmin,* from pg_class where
relname='order_to_do';
oid | ctid | cmax | xmax | cmin | xmin |
relname | relnamespace | reltype | relowner | relam | relfilenode |
relpages | reltuples | reltoastrelid | reltoastidxid | relhasindex |
relisshared | relkind | relnatts | relchecks | reltriggers | relukeys |
relfkeys | relrefs | relhasoids | relhaspkey | relhasrules |
relhassubclass | relacl
----------+---------+------+----------+----------+----------+-------------+--------------+----------+----------+-------+-------------+----------+-----------+---------------+---------------+-------------+-------------+---------+----------+-----------+-------------+----------+----------+---------+------------+------------+-------------+----------------+--------
11462032 | (27,2) | 0 | 2 | 2 | 46034931 |
order_to_do | 2200 | 11462033 | 101 | 0 | 11462032
| 506 | 59401 | 11462039 | 0 | t |
f | r | 5 | 0 | 1 | 0
| 0 | 0 | t | t | f |
f |
11462032 | (27,43) | 2 | 46034931 | 46034931 | 8051642 |
order_to_do | 2200 | 11462033 | 101 | 0 | 11462032
| 453 | 53407 | 11462039 | 0 | t |
f | r | 5 | 0 | 1 | 0
| 0 | 0 | t | t | f |
f |
(2 rows)

>Also, would you check whether you get the same results from this query
>with enable_indexscan switched off?
>
>
I am pretty sure that they are the same with index scan turned off, but
here is the result for verification.

live=# set enable_indexscan=off;
SET
live=# select oid,ctid,cmax,xmax,cmin,xmin,* from pg_class where
relname='order_to_do';
oid | ctid | cmax | xmax | cmin | xmin |
relname | relnamespace | reltype | relowner | relam | relfilenode |
relpages | reltuples | reltoastrelid | reltoastidxid | relhasindex |
relisshared | relkind | relnatts | relchecks | reltriggers | relukeys |
relfkeys | relrefs | relhasoids | relhaspkey | relhasrules |
relhassubclass | relacl
----------+---------+------+----------+----------+----------+-------------+--------------+----------+----------+-------+-------------+----------+-----------+---------------+---------------+-------------+-------------+---------+----------+-----------+-------------+----------+----------+---------+------------+------------+-------------+----------------+--------
11462032 | (27,2) | 0 | 2 | 2 | 46034931 |
order_to_do | 2200 | 11462033 | 101 | 0 | 11462032
| 506 | 59401 | 11462039 | 0 | t |
f | r | 5 | 0 | 1 | 0
| 0 | 0 | t | t | f |
f |
11462032 | (27,43) | 2 | 46034931 | 46034931 | 8051642 |
order_to_do | 2200 | 11462033 | 101 | 0 | 11462032
| 453 | 53407 | 11462039 | 0 | t |
f | r | 5 | 0 | 1 | 0
| 0 | 0 | t | t | f |
f |
(2 rows)

thanks
Jared Carr