Bug report

From: Andrea Baldoni <abaldoni(at)tiscalinet(dot)it>
To: pgsql-bugs(at)postgresql(dot)org
Subject: Bug report
Date: 2000-11-20 15:01:25
Message-ID: 20001120160125.A27827@xcal.net
Views: Raw Message | Whole Thread | Download mbox | Resend email
Thread:
Lists: pgsql-bugs

To report any other bug, fill out the form below and e-mail it to
pgsql-bugs(at)postgresql(dot)org(dot)

============================================================================
POSTGRESQL BUG REPORT TEMPLATE
============================================================================

Your name : Andrea Baldoni
Your email address : abaldoni(at)tiscalinet(dot)it

System Configuration
---------------------
Architecture : Intel Pentium III 500

Operating System : Linux 2.4.0-test10 (same problem on 2.2.17),
debian 2.2

PostgreSQL version : PostgreSQL-6.5.2

Compiler used : gcc version 2.95.2 20000313

Please enter a FULL description of your problem:
------------------------------------------------

Creating tables (temp or normal) inside a transaction sometime fails,
ending with corruption in pg_attribute system data table (if I do
"select * from pg_attribute" the server crash). If I stop and restart the
server the problem remains... I must restore pg_attribute* files to make
things working. The file having the name of the table created sometime
remains in the file system.

Debug output:

.
.
pq_flush: send() failed: Bad file descriptor
pq_flush: send() failed: Bad file descriptor
pq_recvbuf: recv() failed: Bad file descriptor
ERROR: infinite recursion in proc_exit
pq_flush: send() failed: Bad file descriptor
NOTICE: AbortTransaction and not in in-progress state
pq_flush: send() failed: Bad file descriptor
pq_flush: send() failed: Bad file descriptor
pq_recvbuf: recv() failed: Bad file descriptor
ERROR: infinite recursion in proc_exit
.
.

Please describe a way to repeat the problem. Please try to provide a
concise reproducible example, if at all possible:
----------------------------------------------------------------------

I'm using many script like this one, accessing different tables, and I haven't
found differences.
Sample bash script:

#!/bin/bash
umask 077
cat << EOF |psql abaldoni 2> /dev/null > /tmp/$0$$

begin transaction;

create temp table excontitmp0 (
id_documento integer,
totale real
);

create temp table excontitmp1 (
id_documento integer,
totale real
);

create temp table excontitmp2 (
numerario integer
);

insert into excontitmp2 select numerario from conti_desc where
capoconto=$1 and sottoconto=$2;

insert into excontitmp0 select id_documento,
sum(importo) as totale from conti where capo_da=$1 and sotto_da=$2
and documenti.data>='01/01/2000' and
id_documento=documenti.id group by id_documento;

insert into excontitmp0 select id_documento,
-sum(importo) as totale from conti where capo_a=$1 and sotto_a=$2
and documenti.data>='01/01/2000' and
id_documento=documenti.id group by id_documento;

insert into excontitmp1 select id_documento, sum(totale)
as totale from excontitmp0 group by id_documento;

select id_documento, documenti.data, ditte.ragione_sociale,
totale*(1-excontitmp2.numerario*2) from
excontitmp1 where ditte.id=documenti.id_clifor and
documenti.id=id_documento order by documenti.data, id_documento;

select sum(totale*(1-excontitmp2.numerario*2)) as totale from excontitmp0;

commit transaction;

EOF

less /tmp/$0$$
rm /tmp/$0$$

If you know how this problem might be fixed, list the solution below:
---------------------------------------------------------------------

Ciao,
AB

Responses

Browse pgsql-bugs by date

  From Date Subject
Next Message Mike Schulte 2000-11-20 15:48:42 Comparison of Strings
Previous Message Anders Bengtsson 2000-11-18 16:46:10 Re: [INTERFACES] Re: no meaningful way to determine fe or beversion?