Re: duplicate key

At 10:53 -0500 13/01/2003, Tom Lane wrote:
>pascal =?iso-8859-1?Q?b=E9rest?= <lists(at)magnetophone(dot)com> writes:
> > We have exactly the same things in the base.
> > Actually, we have 221 duplicate keys in the table (65000 records)
>
>Seems like a bug, but there's no chance of diagnosing it without more
>info. For starters, what Postgres version are you running?

We are using postgres 7.2.3

> Are all of
>the records with duplicate keys exact duplicates in all other fields
>too?

Yes. When we print all the fields, we have exactly the same entries.
So we can't delete one the two entries. That's quite strange.

>Also, it would be useful to see the ctid, xmin, xmax, cmin,
>and cmax values for some of the sets of duplicate rows.

Could you explain where we can have these informations ?

We also delete the duplicate keys, but we still have some of them.
We delete again and again, and we still have some of them. Finally,
we delete them manually, and it was possible to REINDEX the table.

This problem seams to be important, as we can't do REINDEX or VACUUM
on the database. In decembre 20th, we had have 1.8 Gb database and
now, it's about 5.0 Gb database. Our server is running slowly now.

We didn't have this problem with postgres 7.2.1.

Thanx for your help,
Pascal.