Difference between revisions of "PostgreSQL"

Revision as of 10:37, 7 March 2014

zh-CN:PostgreSQL

PostgreSQL is an open source, community driven, standard compliant object-relational database system.

This document describes how to set up PostgreSQL. It also describes how to configure PostgreSQL to be accessible from a remote client. If you need help setting up the rest of a web stack, see the LAMP page and follow all of the sections except the one related to MySQL.

Before you start

Several sections have instructions stating "become the postgres user". Commands that should be run as the postgres user are prefixed by [postgres]$ in this article.

Execute the following as root to get a shell as the postgres user:

# su - postgres

If you use sudo, you can use the following:

$ sudo -i -u postgres

The postgres user will automatically be created by installing PostgreSQL.

Installing PostgreSQL

Install postgresql from the official repositories.

Before PostgreSQL can function correctly the database cluster must be initialized by the postgres user. Become the postgres user and run the following commmand:

[postgres]$ initdb --locale en_US.UTF-8 -E UTF8 -D '/var/lib/postgres/data'

Start PostgreSQL and, optionally, add it to the list of daemons that start on system startup:

# systemctl start postgresql
# systemctl enable postgresql

Create your first database/user

Become the postgres user. Add a new database user using the createuser command:

[postgres]$ createuser --interactive

Create a new database over which the above user has read/write privileges using the createdb command (execute this command from your login shell if the database user has the same name as your Linux user, otherwise add -U database-username to the following command):

$ createdb myDatabaseName

Familiarize with PostgreSQL

Access the database shell

Become the postgres user. Start the primary db shell, psql, where you can do all your creation of databases/tables, deletion, set permissions, and run raw SQL commands. Use the "-d" option to connect to the database you created (without specifying a database, psql will try to access a database that matches your username)

[postgres]$ psql -d myDatabaseName

Some helpful commands:

Get help

=> \help

Connect to a particular database

=> \c <database>

List all users and their permission levels

=> \du

Shows summary information about all tables in the current database

=> \dt

exit/quit the psql shell

=> \q or CTRL+d

There are of course many more meta-commands, but these should help you get started.

Optional configuration

Configure PostgreSQL to be accessible from remote hosts

The PostgreSQL database server configuration file is postgresql.conf. This file is located in the data directory of the server, typically /var/lib/postgres/data. This folder also houses the other main config files, including the pg_hba.conf.

Edit the file /var/lib/postgres/data/postgresql.conf. In the connections and authentications section add the listen_addresses line to your needs:

listen_addresses = 'localhost,my_remote_ip_address'

Take a careful look at the other lines.

Host-based authentication is configured in /var/lib/postgres/data/pg_hba.conf. This file controls which hosts are allowed to connect. Note that the defaults allow any local user to connect as any database user including the database superuser. Add a line like the following:

# IPv4 local connections:
host   all   all   my_remote_ip_address/32   md5

where your_desired_ip_address is the IP address of the client.

See the documentation for pg_hba.conf.

After this you should restart the daemon process for the changes to take effect with:

# systemctl restart postgresql

For troubleshooting take a look in the server log file

$ tail /var/log/postgresql.log

Change default data directory

The default directory where all your newly created databases will be stored is /var/lib/postgres/data. To change this, follow these steps:

Create the new directory and make the postgres user its owner:

# mkdir -p /pathto/pgroot/data
# chown -R postgres:postgres /pathto/pgroot

Become the postgres user, and initialize the new cluster:

[postgres]$ initdb -D /pathto/pgroot/data

If enabled, disable postgres.service. Copy /usr/lib/systemd/system/postgresql.service to /etc/systemd/system/postgresql.service and edit it to change the default PGROOT and PIDFile paths.

Environment=PGROOT=/pathto/pgroot/
...
PIDFile=/pathto/pgroot/data/postmaster.pid

Change default encoding of new databases to UTF-8

When creating a new database (e.g. with createdb blog) PostgreSQL actually copies a template database. There are two predefined templates: template0 is vanilla, while template1 is meant as an on-site template changeable by the administrator and is used by default. In order to change the encoding of new database, one of the options is to change on-site template1. To do this, log into PostgresSQL shell (psql) and execute the following:

First, we need to drop template1. Templates cannot be dropped, so we first modify it so it is an ordinary database:

UPDATE pg_database SET datistemplate = FALSE WHERE datname = 'template1';

Now we can drop it:

DROP DATABASE template1;

The next step is to create a new database from template0, with a new default encoding:

CREATE DATABASE template1 WITH TEMPLATE = template0 ENCODING = 'UNICODE';

Now modify template1 so it is actually a template:

UPDATE pg_database SET datistemplate = TRUE WHERE datname = 'template1';

(OPTIONAL) If you do not want anyone connecting to this template, set datallowconn to FALSE:

UPDATE pg_database SET datallowconn = FALSE WHERE datname = 'template1';

Now you can create a new database:

[postgres]$ createdb blog

If you log in back to psql and check the databases, you should see the proper encoding of your new database:

\l

returns

                              List of databases
  Name    |  Owner   | Encoding  | Collation | Ctype |   Access privileges
-----------+----------+-----------+-----------+-------+----------------------
blog      | postgres | UTF8      | C         | C     |
postgres  | postgres | SQL_ASCII | C         | C     |
template0 | postgres | SQL_ASCII | C         | C     | =c/postgres
                                                     : postgres=CTc/postgres
template1 | postgres | UTF8      | C         | C     |

Administration tools

• phpPgAdmin — Web-based administration tool for PostgreSQL.

http://phppgadmin.sourceforge.net || phppgadmin

• pgAdmin — GUI-based administration tool for PostgreSQL.

http://www.pgadmin.org/ || pgadmin3

Upgrading PostgreSQL

Quick guide

This is for upgrading from 9.2 to 9.3.

 pacman -S --needed postgresql-old-upgrade
 su -
 su - postgres -c 'mv /var/lib/postgres/data /var/lib/postgres/data-9.2'
 su - postgres -c 'mkdir /var/lib/postgres/data'
 su - postgres -c 'initdb --locale en_US.UTF-8 -E UTF8 -D /var/lib/postgres/data'

If you had custom settings in configuration files like pg_hba.conf and postgresql.conf, merge them into the new ones. Then:

 su - postgres -c 'pg_upgrade -b /opt/pgsql-9.2/bin/ -B /usr/bin/ -d /var/lib/postgres/data-9.2 -D /var/lib/postgres/data'

If the "pg_upgrade" step fails with:

• cannot write to log file pg_upgrade_internal.log
  Failure, exiting
  Make sure you're in a directory that the "postgres" user has enough rights to write the log file to (/tmp for example). Or use "su - postgres" instead of "sudo -u postgres".
• LC_COLLATE error that says that old and new values are different
  Figure out what the old locale was, C or en_US.UTF-8 for example, and force it when calling initdb.

 sudo -u postgres LC_ALL=C initdb -D /var/lib/postgres/data

• There seems to be a postmaster servicing the old cluster.
  Please shutdown that postmaster and try again.
  Make sure postgres isn't running. If you still get the error then chances are these an old PID file you need to clear out.

 > sudo -u postgres ls -l /var/lib/postgres/data-9.2
   total 88
   -rw------- 1 postgres postgres     4 Mar 25  2012 PG_VERSION
   drwx------ 8 postgres postgres  4096 Jul 17 00:36 base
   drwx------ 2 postgres postgres  4096 Jul 17 00:38 global
   drwx------ 2 postgres postgres  4096 Mar 25  2012 pg_clog
   -rw------- 1 postgres postgres  4476 Mar 25  2012 pg_hba.conf
   -rw------- 1 postgres postgres  1636 Mar 25  2012 pg_ident.conf
   drwx------ 4 postgres postgres  4096 Mar 25  2012 pg_multixact
   drwx------ 2 postgres postgres  4096 Jul 17 00:05 pg_notify
   drwx------ 2 postgres postgres  4096 Mar 25  2012 pg_serial
   drwx------ 2 postgres postgres  4096 Jul 17 00:53 pg_stat_tmp
   drwx------ 2 postgres postgres  4096 Mar 25  2012 pg_subtrans
   drwx------ 2 postgres postgres  4096 Mar 25  2012 pg_tblspc
   drwx------ 2 postgres postgres  4096 Mar 25  2012 pg_twophase
   drwx------ 3 postgres postgres  4096 Mar 25  2012 pg_xlog
   -rw------- 1 postgres postgres 19169 Mar 25  2012 postgresql.conf
   -rw------- 1 postgres postgres    48 Jul 17 00:05 postmaster.opts
   -rw------- 1 postgres postgres    80 Jul 17 00:05 postmaster.pid   # <-- This is the problem
 
 > sudo -u postgres mv /var/lib/postgres/data-9.2/postmaster.pid /tmp

• ERROR: could not access file "$libdir/postgis-2.0": No such file or directory
  Retrieve postgis-2.0.so from postgis package for version postgresql 9.2 () and copy it to /opt/pgsql-9.2/lib (make sure the privileges are right)

Detailed instructions

Note that these instructions could cause data loss. Use at your own risk.

It is recommended to add the following to your /etc/pacman.conf file:

IgnorePkg = postgresql postgresql-libs

This will ensure you do not accidentally upgrade the database to an incompatible version. When an upgrade is available, pacman will notify you that it is skipping the upgrade because of the entry in pacman.conf. Minor version upgrades (e.g., 9.0.3 to 9.0.4) are safe to perform. However, if you do an accidental upgrade to a different major version (e.g., 9.0.X to 9.1.X), you might not be able to access any of your data. Always check the PostgreSQL home page (http://www.postgresql.org/) to be sure of what steps are required for each upgrade. For a bit about why this is the case see the versioning policy.

There are two main ways to upgrade your PostgreSQL database. Read the official documentation for details.

For those wishing to use pg_upgrade, a postgresql-old-upgrade package is available in the repositories that will always run one major version behind the real PostgreSQL package. This can be installed side by side with the new version of PostgreSQL. When you are ready to perform the upgrade, you can do

pacman -Syu postgresql postgresql-libs postgresql-old-upgrade

Note also that the data directory does not change from version to version, so before running pg_upgrade it is necessary to rename your existing data directory and migrate into a new directory. The new database must be initialized, as described near the top of this page.

# systemctl stop postgresql
# su - postgres -c 'mv /var/lib/postgres/data /var/lib/postgres/olddata'
# su - postgres -c 'initdb --locale en_US.UTF-8 -E UTF8 -D /var/lib/postgres/data'

Reference the upstream pg_upgrade documentation for details.

The upgrade invocation will likely look something like the following (run as the postgres user). Do not run this command blindly without understanding what it does!

# su - postgres -c 'pg_upgrade -d /var/lib/postgres/olddata/ -D /var/lib/postgres/data/ -b /opt/pgsql-8.4/bin/ -B /usr/bin/'

Manual dump and reload

You could also do something like this (after the upgrade and install of postgresql-old-upgrade) (NB: below is command for postgres8.4 update, you can find similar command in /opt/ for postgres 9.2 update. )

# systemctl stop postgresql
# /opt/pgsql-8.4/bin/pg_ctl -D /var/lib/postgres/olddata/ start
# /opt/pgsql-8.4/bin/pg_dumpall >> old_backup.sql
# /opt/pgsql-8.4/bin/pg_ctl -D /var/lib/postgres/olddata/ stop
# systemctl start postgresql
# psql -f old_backup.sql postgres

Troubleshooting

Improve performance of small transactions

If you are using PostgresSQL on a local machine for development and it seems slow, you could try turning synchronous_commit off in the configuration (/var/lib/postgres/data/postgresql.conf). Beware of the caveats, however.

synchronous_commit = off

Prevent disk writes when idle

PostgreSQL periodically updates its internal "statistics" file. By default, this file is stored on disk, which prevents disks spinning down on laptops and causes hard drive seek noise. It's simple and safe to relocate this file to a memory-only file system with the following configuration option:

stats_temp_directory = '/run/postgresql'