Message-ID: <0c1f6b24a56b4ee79331c656efce9ca622ffa0a1.camel@cybertec.at>
Subject: Re: Problem with constraint unique.
From: Laurenz Albe <laurenz.albe@cybertec.at>
To: =?UTF-8?Q?=D0=9C=D0=B0=D1=80=D0=B0=D1=82_?=
 =?UTF-8?Q?=D0=93=D0=B0=D1=81=D0=B0=D0=BD=D1=8F=D0=BD?=
	 <marat_gasanyan@mail.ru>, pgsql-bugs@lists.postgresql.org
Date: Thu, 12 Dec 2024 07:47:24 +0100
In-Reply-To: <1733928443.950047988@f125.i.mail.ru>
References: <1733928443.950047988@f125.i.mail.ru>
Content-Type: text/plain; charset="UTF-8"
Content-Transfer-Encoding: quoted-printable
User-Agent: Evolution 3.54.2 (3.54.2-1.fc41) 
MIME-Version: 1.0
Archived-At: 
 <https://www.postgresql.org/message-id/0c1f6b24a56b4ee79331c656efce9ca622ffa0a1.camel%40cybertec.at>
Precedence: bulk

On Wed, 2024-12-11 at 17:47 +0300, =D0=9C=D0=B0=D1=80=D0=B0=D1=82 =D0=93=D0=
=B0=D1=81=D0=B0=D0=BD=D1=8F=D0=BD wrote:
> I apologies for bothering your team,=C2=A0before send this message we goo=
gle the
> problem and was not able to find any tips about it.=C2=A0Our project enco=
unter
> with quite strange situation where we found violation of unique constrain=
t
> of primary key. We have a table that has an attribute ID as PK. The table
> name enerstorymain_task_info. Surprisingly the query=C2=A0
> select id, count(*) from=C2=A0=C2=A0enerstroymain_task_info group by 1 ha=
ving count(*) > 1
> return a lot of rows. Than we rechecked PK it is ok. So we supposed that
> some how we had dropped the PK and then created duplicates and than rebui=
ld
> the PK, but in this case PK would not rebuild. To avoid any weird setting=
s
> I run the query from PGAdmin and DBView from different computers.
> Also I doubt that any kind problem like repeatable red could=C2=A0be the =
reason
> due to insert operation is done only a server without any complicated que=
ry.
> I mean simple INSERT INTO =E2=80=A6.

No, dropping and re-creating the constraint cannot be the explanation,
unless the primary key constraint is NOT VALID.

With "uuid", it also cannot be a collation problem.

The best thing to get rid of this data corruption is to delete the extra
entries and then to sump and restore the data to a new database.

Check if your hardware has problems.

Yours,
Laurenz Albe