SELECT fails to present result rows depending on the columns to show

On Tue, Nov 23, 2021 at 09:15:06AM +0100, Matthias Apitz wrote:

Hello,

We encounter the following problem in a 13.1 server on Linux:

sisis=# select desk, feldnr from titel_worte where desk = '2' and feldnr = 257;
desk | feldnr
------+--------
2 | 257
(1 row)

but:

sisis=# select * from titel_worte where desk = '2' and feldnr = 257;
desknr | feldnr | desk | deskorg | gesanz | aufanz | katkey1 | katkey2
--------+--------+------+---------+--------+--------+---------+---------
(0 row)

sisis=# select desk, feldnr, deskorg from titel_worte where desk = '2' and feldnr = 257;
desk | feldnr | deskorg
------+--------+---------
(0 row)

The table was created as:

create table titel_worte (
desknr serial ,
feldnr SMALLINT ,
desk VARCHAR (245) ,
deskorg VARCHAR (245) ,
gesanz INTEGER ,
aufanz INTEGER ,
katkey1 INTEGER ,
katkey2 INTEGER
)
;

There are no messages in the serverlog when the SELECT fails to show the
row.

What could be the reason for this? Thanks

Broken index could. Then this anomaly shoud have gone after reindex table.

Ilya Anfimov schrieb am 23.11.2021 um 09:31:

but:

sisis=# select * from titel_worte where desk = '2' and feldnr = 257;
desknr | feldnr | desk | deskorg | gesanz | aufanz | katkey1 | katkey2
--------+--------+------+---------+--------+--------+---------+---------
(0 row)

sisis=# select desk, feldnr, deskorg from titel_worte where desk = '2' and feldnr = 257;
desk | feldnr | deskorg
------+--------+---------
(0 row)

The table was created as:

create table titel_worte (
desknr serial ,
feldnr SMALLINT ,
desk VARCHAR (245) ,
deskorg VARCHAR (245) ,
gesanz INTEGER ,
aufanz INTEGER ,
katkey1 INTEGER ,
katkey2 INTEGER
)
;

There are no messages in the serverlog when the SELECT fails to show the
row.

What could be the reason for this? Thanks

Broken index could. Then this anomaly shoud have gone after reindex table.

Ilya refers to the problems decribed here:

https://wiki.postgresql.org/wiki/Locale_data_changes

El día martes, noviembre 23, 2021 a las 10:09:36 +0100, Thomas Kellerer escribió:

Broken index could. Then this anomaly shoud have gone after reindex table.

Ilya refers to the problems decribed here:

https://wiki.postgresql.org/wiki/Locale_data_changes

Thanks for the pointer. What is written there matches with the system
patch by our IT department some days ago to all our SLES 15 servers.

We have to talk and see what we have todo.

matthias