Documentation error in postgresql-common

[Shelby Cain]

Binary package hint: postgresql-common

The file /etc/cron.d/postgresql-common installed by the postgresql-common package offers the following advice:

# On Sunday you may wish to run a VACUUM FULL ANALYSE as well
# If you do not run a 24/7 site, you may want to uncomment the next line
# so as to do a regular VACUUM FULL. If you need 24/7 connectivity, save
# VACUUM FULL for when you think you really need it.
# 10 3 * * Sun root if [ -x /usr/sbin/pg_maintenance ]; then /usr/sbin/pg_maintenance --full --analyze >/dev/null; fi

I believe the general consensus of the Postgresql community is that running "vacuum full" periodically against a database is rarely considered a good idea as it can lead to issues like index bloat. Perhaps this file should reflect that sentiment and point interested people to the documentation on Postgresql's website so that they can fully understand the impact that "vacuum full" can have on a database.

[Shelby Cain (alyandon)]

This is on an Ubuntu 7.10 system by the way.

[Martin Pitt (pitti)]

Fixed in bzr head.

[Martin Pitt (pitti)]

 postgresql-common (84) unstable; urgency=low
 .
   * Add t/140_pg_config.t: Check correct output of pg_config, to avoid bugs
     like #455509 and #462037 in the future.
   * t/040_upgrade.t: Create and cd into an inaccessible test directory.
   * pg_upgradecluster: Chdir to / before the upgrade to avoid confusing error
     messages about inaccessible cwd.
   * debian/README.Debian: Update for 8.3 being the default version. Drop
     description of 7.4 autovacuum daemon magic, since it is obsolete now.
   * debian/postgresql-common.templates: Point to README.Debian.
   * debian/postgresql-common.cron.d: Update for 8.3, and add warnings and doc
     pointer about enabling regular VACUUM FULL. (LP: #186831)
   * PgCommon.pm, get_cluster_locales(): Print proper error message when server
     package is removed, but not purged, and a cluster still exists.
     (Closes: #463070)
   * t/030_errors.t, t/130_nonroot_admin.t: Check pg_dropcluster/
     pg_createcluster error message when being called with insufficient
     privileges.
   * pg_createcluster, pg_dropcluster: Point out that program needs to be run
     as root when creating/removing config or data directory fails.
     (Closes: #460576)
   * t/090_multicluster.t: Remove test case for nonexisting PGSERVICE value,
     since this is not reliably reported in psql.
   * t/140_pg_config.t: Skip pg_config test for 8.1. It's known broken in Etch
     (pg_config was only shipped in libpq-def) and irrelevant in Lenny.