Home/ Questions/Q 7989663
In Process

The Archive Base Latest Questions

Editorial Team
  • 0
Asked: 2026-06-04T12:49:01+00:00

I have been working to upgrade a Postgresql database from 8.4 to 9.1. I

  • 0

I have been working to upgrade a Postgresql database from 8.4 to 9.1. I have encountered innumerable problems (may have to do with Ubuntu’s file placements), but I think I see the finish line.
Pg_upgrade is almost working, but it fails when it reaches the first database, ‘template0,’ because it ‘already exists.’ 

psql:/var/lib/postgresql/pg_upgrade_dump_globals.sql:36: ERROR:  database "template0" already exists

There were problems executing "/usr/lib/postgresql/9.1/bin/psql" --set ON_ERROR_STOP=on --no-psqlrc --port 5432 --username "postgres" -f "/var/lib/postgresql/pg_upgrade_dump_globals.sql" --dbname template1 >> "/dev/null"
Failure, exiting

This template0 database is of course part of every new cluster and can not be deleted. I tried renaming template0 and template1 (to be safe) on the new cluster (after recreating it), but the subsequent pg_upgrade call also failed, but this time complaining that template1 didn’t exist on the new cluster. 

"/usr/lib/postgresql/9.1/bin/pg_ctl" -w -l "/dev/null" -D "/var/lib/postgresql/9.1/main" -o "-p 5432 -b" start >> "/dev/null" 2>&1

connection to database failed: FATAL:  database "template1" does not exist

unable to connect to new postmaster started with the command: "/usr/lib/postgresql/9.1/bin/pg_ctl" -w -l "/dev/null" -D "/var/lib/postgresql/9.1/main" -o "-p 5432 -b" start >> "/dev/null" 2>&1
Failure, exiting

I tried it again only renaming template0 and it failed again, but this time with some assertion issue from postgres itself.

executing: SELECT datcollate, datctype FROM     pg_catalog.pg_database WHERE    datname = 'template0' 

pg_upgrade: /build/buildd/postgresql-9.1-9.1.3/build/../contrib/pg_upgrade/check.c:310: set_locale_and_encoding: Assertion `PQntuples(res) == 1' failed.
Aborted (core dumped)

Sanity Check Notes: In case anyone thinks ‘just do a full sql dump and restore it’ that isn’t working well either, which is why I am trying to get pg_upgrade working. I also made sure to drop and recreate the cluster every time an upgrade failed to prevent any relics of failures past from continuing to screw with the upgrade. The database is on an Amazon Volume, so I can’t really break anything and always have a backup. I am on ubuntu 12 and have both databases installed side by side.

I am currently out of ideas that doesn’t involve going into the postgresql source and would greatly appreciate any help and suggestions as to what to try next. Thanks in advance.

Share

Leave an answer

You must login to add an answer.

Need An Account,

1 Answer

  1. Editorial Team

    Stop. Make a cup of tea / coffee and sit down.

    1. Make sure you have 9.1 and 8.4 installed at the same time. Debian allows this, so presumably ubuntu does too.
    2. Using the 9.1 version of pg_dump, dump each of your databases from 8.4, and either a dumpall for the users or just manually recreate them.
    3. Using 9.1 pg_restore, restore your databases as user postgres along with the –create flag.

    If there are specific problems with any of these steps you’ll need to make a note of them and provide details. Oh – the best place for detailed step-by-step advice is probably the postgresql-general mailing list (see the website for details). You can take your time there and get each step right.

    At a guess you’re trying to do the pg_upgrade into an existing initdb-ed directory which I don’t think it expects. In any case, dump+restore is the cleanest way to do this – it also gives you a backup copy of your DB too.

    • 0