Re: pg_dump problem - Mailing list pgsql-general

From Tom Lane
Subject Re: pg_dump problem
Date
Msg-id 12446.1219939625@sss.pgh.pa.us
Whole thread Raw
In response to pg_dump problem  ("Laura Del Caño" <ldelcano@gmail.com>)
Responses Re: pg_dump problem  ("Laura Del Caño" <ldelcano@gmail.com>)
List pgsql-general
"=?ISO-8859-1?Q?Laura_Del_Ca=F1o?=" <ldelcano@gmail.com> writes:
> I am issuing the following command:
> pg_dump -c -o -s -n distributed -f backups/schema.sql syslog
> which apparently works fine. It is only when I tried to restore it in
> a fresh created database using:
> psql syslog < backups/schema.sql
> that I get lots of errors saying objects do not exist.

> Having a look at the schema.sql file I see the following:

> SET search_path = distributed, pg_catalog;
> ...
> CREATE SCHEMA distributed;

Is that SET located before/during the sequence of DROPs issued due to -c ?

I think this is a known bug in the behavior of -c.  It was fixed before
8.3 release but doesn't seem to have been back-patched into any older
branches:


http://anoncvs.postgresql.org/cvsweb.cgi/pgsql/src/bin/pg_dump/pg_backup_archiver.c.diff?r1=text&tr1=1.149&r2=text&tr2=1.151

Simplest workaround with older pg_dumps might be to avoid -c and
just do a manual "DROP SCHEMA distributed CASCADE" before restoring
the dump.

            regards, tom lane

pgsql-general by date:

Previous
From: Bill Moran
Date:
Subject: Re: Vaccuum best practice: cronjob or autovaccuum?
Next
From: Joao Ferreira gmail
Date:
Subject: Re: Vaccuum best practice: cronjob or autovaccuum?