Re: postgres 7.2.1 pg_restore large objects - Mailing list pgsql-admin

From Dario
Subject Re: postgres 7.2.1 pg_restore large objects
Date
Msg-id MHEDJHCKDNOEHJKHIOCJAEGPCDAA.dario_d_s@unitech.com.ar
Whole thread Raw
In response to Re: postgres 7.2.1 pg_restore large objects  (Juan Miguel <juanmime@ono.com>)
List pgsql-admin
Regarding this old subject, we moved to 8.0.2. Updating to 7.2.5 didn't
solved the issue (I could not find an 7.2.7 rpm release for Red hat 7.2.
since that's the plataform already installed by client, I had to stick to
that). The upgrade to 8.0.2 was done in a Red Hat 9.0 system.

Import of 'lo' was now succesfully.

Greetings


---
I'll resume here steps taken to upgrade(not recommended, just taken for
compatibility with a VB6 app with ODBC)
-----------------------------------------------------
usefull changes in postgresql.conf
regex_flavor = advanced    # advanced, extended, or basic
default_with_oids = true
-----------------------------------------------------
createdb --encoding SQL-ANSI BASEDEDATOS               (for import from 7.2
database)
-----------------------------------------------------
our app use this datatype:
CREATE DOMAIN datetime AS timestamp
---------------------------------------------------------------
From the docs: An empty string ('') is no longer allowed as the input into
an integer field. Formerly, it was silently interpreted as 0.

This kind of insertion:
INSERT INTO rca_crt (order) VALUES ('')
had to be changed.


-----Mensaje original-----
De: pgsql-admin-owner@postgresql.org
[mailto:pgsql-admin-owner@postgresql.org]En nombre de Juan Miguel
Enviado el: domingo, 03 de abril de 2005 7:44
Para: Dario Pudlo
CC: pgsql-admin@postgresql.org
Asunto: Re: [ADMIN] postgres 7.2.1 pg_restore large objects


Dario Pudlo wrote:
Problem: Not being able to restore a database with lo. I have already read
forums and doc. Probably I'm loosing something, but I just can figure.

Reading forums I though I finally got a "right" way to do it.
I'm running this as OS user "postgres" which is a super user

pg_dump -Ft -o -b base > base.tar 2>log_err2 # If I untar the file, i see
table data, and blob data, toc and restore.sql.

createdb base2 //from OS shell

/usr/local/pgsql/bin/pg_restore -Ft -s -v -d base2 base.tar  2>elogerr
#creates schema

psql -f src/contrib/lo.sql base2    #creates lo type and functions... my
base.tar includes lo type definition. O also tried to delete the type
previously, and did not make any difference. If I do not do this, pg_restore
claims that I lo and oid must be explicity casted. One more thing, I can't
make lo_manage to work (trigger type doesn't exist). I don use it anyway,
since I manage lo updates and delete from app. I delete blobs from app.

/usr/local/pgsql/bin/pg_restore -Ft -a -v -d base2 base.tar 2>flogerr

the pg_restore fails with the following error.

----------------------------------
pg_restore log (flogerr file)
pg_restore: [archiver (db)] error while updating column "tca_texto" of table
"tca": server closed the connection unexpectedly
        This probably means the server terminated abnormally
        before or while processing the request.
pg_restore: *** aborted because of error

----------------------------------
postgres log file
DEBUG:  server process (pid 25276) was terminated by signal 11
DEBUG:  terminating any other active server processes
NOTICE:  Message from PostgreSQL backend:
        The Postmaster has informed me that some other backend
        died abnormally and possibly corrupted shared memory.
        I have rolled back the current transaction and am
        going to terminate your database system connection and exit.
        Please reconnect to the database system and repeat your query.
DEBUG:  all server processes terminated; reinitializing shared memory and
semaphores
----------------------------------

No table is updated.

I tried to create a table with oid column instead of lo, restore (worked in
this way), I renamed table, I created a table with lo datatype, but when I
tried to insert into ... , same error occurs.

server closed the connection unexpectedly
This probably means the server terminated abnormally
before or while processing the request.


I have two different versions of postgres, one compiled by someone here, and
another version, installed via rpm.
unlikely, upgrading to a different version, is not an option, I probably
update to 7.2.(last version), but only if this specific problem is solved. I
have read the release section of the doc and did not apeared explicity...

greetings.

TIA to all

"Hoy es un buen dia para morir" (today is a good day to die, Klingon en
espanol)!


---------------------------(end of broadcast)---------------------------
TIP 7: don't forget to increase your free space map settings


Hello, I use lo type in several databases. I use postgres since version 6.4,
and I started to use "lo" type since version 7.x.
Well, I remember that initially restoring tables with lo type columns made
me headaches in early 7.x versions. Nowaydays I only have a database 7.2
with lo types. I dump in 7.2, and then I restore a copy in other machine in
7.5. This is the process that I use to dump and restore the database.

In 7.2 (the production system):

1. I dump the database.

pg_dump -Ft -b -o -U postgres databasename > dbdump_date.tar

Then in the testing machine (7.5) I do this steps:

1. Create de database

createdatabase -U dbusername databasename

2.- Restore only the structure

/usr/local/pgsql/bin/pg_restore -Ft -s -v -U postgres -d databasename
dbdump_date.tar

Why I restore the structure like postgres user ? Because, the dump file,
restores the lo type in the database (only the super user can do that),
after that changes to the correct user (dbusername) for creating the
structure of the database.

3.- Restore Only the data, "without triggers"

/usr/local/pgsql/bin/pg_restore -Ft -a -v -X disable-triggers -U postgres -d
databasename dbdump_date.tar


This is the method that I use for dumping and restoring, and works fine for
me. I have serveral databases with lo types in 7.4 and 7.5, and dumping and
restoring in these versions works fine for me (in those cases I restore the
full database in one step, structure and data at same time). I only need to
restore first structure and after data without triggers, when the dump is
from 7.2 version.

Phearps are missing some parameters in the restore and dumping process, but
basically are that I have written.

I hope my experience will be usefull for you.


pgsql-admin by date:

Previous
From: "Margabandhu De.Sambath"
Date:
Subject: unregister
Next
From: "Marc G. Fournier"
Date:
Subject: Network Storage Devices / NFS ...