PostgreSQL

来自 Arch Linux 中文维基

PostgreSQL是一个开源的,社区驱动的,符合标准的 对象-关系型 数据库系统。

安装[编辑 | 编辑源代码]

警告: 在安装新版本的 PostgreSQL 包前,请先查看 #Upgrading PostgreSQL 一节中的必要步骤。

安装 postgresql,这同时会创建一个名为 postgres 的新系统用户。

你现在可以通过提权工具来切换到 postgres 用户下。

注意: 在本篇文章中需要以 postgres 用户运行的命令以 [postgres]$ 作为前置符号。

你可以通过下列命令来切换到 PostgreSQL 用户下:

  • 如果安装了 sudo 且当前用户属于 sudoers
    $ sudo -iu postgres
  • 其他情况下使用 su
    $ su
    # su -l postgres
    

详细用法请参考 sudo(8)su(1)

初始化配置[编辑 | 编辑源代码]

在可以正常使用 PostgreSQL 之前,数据库集群必须被初始化:

[postgres]$ initdb -D /var/lib/postgres/data

其中 -D 提供了数据库集群的默认数据存放位置(如果需要修改目录位置,可以参考#修改默认数据目录)。initdb 也支持多种其它的命令行参数:

这篇文章的某些内容需要扩充。

原因: PostgreSQL also supports ICU locales.[1] (在 Talk:PostgreSQL 中讨论)
  • By default, the locale and the encoding for the database cluster are derived from your current environment (using $LANG value). If this is not what you want, you can override the defaults using --locale=locale (where locale is to be chosen amongst the system's available locales) and --encoding=encoding (which must match the chosen locale). (Once the database is up, you can check which values were used with [postgres]$ psql -l.)
    注意: Using a locale other than C.UTF-8, C, POSIX or ucs_basic can result in a collation version mismatch that will require reindexing if the library providing the locale (glibc or icu) gets updated. See FS#77445.
  • 如果数据目录所在文件系统没有数据校验和功能,你可能会想要启用 PostgreSQL 自带的校验和功能来提高数据完整性保障 - 使用 --data-checksums 参数即可。相关细节可参考 #Enable data checksumming。(Once the database is up, you can check if it is enabled with [postgres]$ psql -c "SHOW data_checksums".)
  • The trust authentication method is used by default, meaning that anyone on the host can connect as any database user. You can use --auth-local=peer --auth-host=scram-sha-256 for safer authentication methods.
  • For more options, see initdb --help and official documentation.

Example:

[postgres]$ initdb --locale=C.UTF-8 --encoding=UTF8 -D /var/lib/postgres/data --data-checksums

Many lines should now appear on the screen with several ending by ... ok:

The files belonging to this database system will be owned by user "postgres".
This user must also own the server process.

The database cluster will be initialized with locale "C.UTF-8".
The default text search configuration will be set to "english".

Data page checksums are enabled.

creating directory /var/lib/postgres/data ... ok
creating subdirectories ... ok
selecting dynamic shared memory implementation ... posix
selecting default max_connections ... 100
selecting default shared_buffers ... 128MB
selecting default time zone ... UTC
creating configuration files ... ok
running bootstrap script ... ok
performing post-bootstrap initialization ... ok
syncing data to disk ... ok

initdb: warning: enabling "trust" authentication for local connections
initdb: hint: You can change this by editing pg_hba.conf or using the option -A, or --auth-local and --auth-host, the next time you run initdb.

Success. You can now start the database server using:

    pg_ctl -D /var/lib/postgres/data -l logfile start

If these are the kind of lines you see, then the process succeeded. Return to the regular user using exit.

注意: To read more about this WARNING, see #Restricts access rights to the database superuser by default.
提示:If you change the root to something other than /var/lib/postgres, you will have to edit the service file. If the root is under home, make sure to set ProtectHome to false.
警告:
  • 如果数据库位于Btrfs文件系统上,你应该在创建数据库前禁用数据库目录的Copy-on-Write
  • 如果数据库位于 ZFS 文件系统上,你应该在创建数据库前先查阅 ZFS#Databases

最后,启动启用 postgresql.service 服务。

创建第一个数据库/用户[编辑 | 编辑源代码]

提示:如果创建一个与你的 Arch 用户 ($USER) 同名的数据库用户,并允许其访问 PostgreSQL 数据库的 shell,那么在使用PostgreSQL 数据库 shell 的时候无需指定用户登录(这样做会比较方便)。

以 postgres 用户身份, 使用 createuser 命令添加一个新的数据库用户:

[postgres]$ createuser --interactive

使用 createdb 命令,创建一个上述用户可以读写的新数据库(execute this command from your login shell if the database user has the same name as your Linux user, otherwise add -O database-username to the following command):

$ createdb myDatabaseName
提示:If you did not grant your new user database creation privileges, add -U postgres to the previous command.

熟悉 PostgreSQL[编辑 | 编辑源代码]

连接数据库 shell[编辑 | 编辑源代码]

登录为 postgres 用户,启动主数据库 shell psql,你可以创建/删除数据库或表、配置权限和运行 SQL 命令。使用 -d 选项连接你创建的数据库(如果没有指定数据库,psql 会尝试连接与你用户名同名的数据库)。

[postgres]$ psql -d myDatabaseName

一些有用的命令:

  • 获取帮助
=> \help
  • 列出所有数据库
=> \l
  • 连接到特定数据库
=> \c database
  • 列出所有用户以及他们的权限
=> \du
  • 展示当前数据库中所有的表相关的汇总信息
=> \dt
  • 退出 psql
=> \q 或是 Ctrl+d

当然也有更多元命令,但这些应该能够帮助您开始。要查看所有元命令:

=> \?

可选配置[编辑 | 编辑源代码]

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 configuration files, including the pg_hba.conf which defines authentication settings, for both local users and other hosts ones.

注意: By default, this folder will not be browsable or searchable by a regular user. This is why find and locate are not finding the configuration files.

Restricts access rights to the database superuser by default[编辑 | 编辑源代码]

The defaults pg_hba.conf allow any local user to connect as any database user, including the database superuser. This is likely not what you want, so in order to restrict global access to the postgres user, change the following line:

/var/lib/postgres/data/pg_hba.conf
# TYPE  DATABASE        USER            ADDRESS                 METHOD

# "local" is for Unix domain socket connections only
local   all             all                                     trust

To:

/var/lib/postgres/data/pg_hba.conf
# TYPE  DATABASE        USER            ADDRESS                 METHOD

# "local" is for Unix domain socket connections only
local   all             postgres                                peer

You might later add additional lines depending on your needs or software ones.

Require password for login[编辑 | 编辑源代码]

Edit /var/lib/postgres/data/pg_hba.conf and set the authentication method for each user (or "all" to affect all users) to scram-sha-256 (preferred), or md5 (less secure; should be avoided if possible):

/var/lib/postgres/data/pg_hba.conf
# TYPE  DATABASE        USER            ADDRESS                 METHOD                                                                                               
                                                                                                                                                                                                                                                              
# "local" is for Unix domain socket connections only                                                                                                                        
local   all             user                                    scram-sha-256

If you choose scram-sha-256, you must also edit /var/lib/postgres/data/postgresql.conf and set:

/var/lib/postgres/data/postgresql.conf
password_encryption = scram-sha-256

Restart postgresql.service, and then re-add each user's password using ALTER USER user WITH ENCRYPTED PASSWORD 'password';.

Configure PostgreSQL to be accessible exclusively through UNIX Sockets[编辑 | 编辑源代码]

In the connections and authentications section of your configuration, set:

/var/lib/postgres/data/postgresql.conf
listen_addresses = ''

This will disable network listening completely. After this you should restart postgresql.service for the changes to take effect.

Configure PostgreSQL to be accessible from remote hosts[编辑 | 编辑源代码]

In the connections and authentications section, set the listen_addresses line to your needs:

/var/lib/postgres/data/postgresql.conf
listen_addresses = 'localhost,my_local_ip_address'

You can use '*' to listen on all available addresses.

注意: PostgreSQL uses TCP port 5432 by default for remote connections. Make sure this port is open in your firewall and able to receive incoming connections. You can also change it in the configuration file, right below listen_addresses

Then add a line like the following to the authentication config:

/var/lib/postgres/data/pg_hba.conf
# TYPE  DATABASE        USER            ADDRESS                 METHOD
# IPv4 local connections:
host    all             all             ip_address/32   md5

where ip_address is the IP address of the remote client.

See the documentation for pg_hba.conf.

注意: Neither sending your plain password nor the md5 hash (used in the example above) over the Internet is secure if it is not done over an SSL-secured connection. See Secure TCP/IP Connections with SSL for how to configure PostgreSQL with SSL.

After this you should restart postgresql.service for the changes to take effect.

For troubleshooting take a look in the server log file:

# journalctl -u postgresql.service

Configure PostgreSQL authenticate against PAM[编辑 | 编辑源代码]

PostgreSQL offers a number of authentication methods. If you would like to allow users to authenticate with their system password, additional steps are necessary. First you need to enable PAM for the connection.

For example, the same configuration as above, but with PAM enabled:

/var/lib/postgres/data/pg_hba.conf
# IPv4 local connections:
host   all   all   my_remote_client_ip_address/32   pam

The PostgreSQL server is however running without root privileges and will not be able to access /etc/shadow. We can work around that by allowing the postgres group to access this file:

# setfacl -m g:postgres:r /etc/shadow

修改默认数据目录[编辑 | 编辑源代码]

默认设置下,新创建的数据库会被存放于 /var/lib/postgres/data 目录下。如果要更改目录位置,可以参考下列步骤:

创建一个新文件夹,并将其所有者设为 postgres 用户:

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

切换到 postgres 用户,然后初始化新集群:

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

通过编辑 postgresql.service附加配置片段,以覆盖 EnvironmentPIDFile 设置。例如:

/etc/systemd/system/postgresql.service.d/PGROOT.conf
[Service]
Environment=PGROOT=/pathto/pgroot
PIDFile=/pathto/pgroot/data/postmaster.pid

如果你想使用 /home 目录作为默认目录或用于表空间,需要在此文件中额外添加一行:

ProtectHome=false

Change default encoding of new databases to UTF-8[编辑 | 编辑源代码]

注意: If you ran initdb with -E UTF8 or while using an UTF-8 locale, these steps are not required.

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 a new database, one of the options is to change on-site template1. To do this, log into PostgreSQL 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';

Optionally, if you do not want anyone connecting to this template, set datallowconn to FALSE:

UPDATE pg_database SET datallowconn = FALSE WHERE datname = 'template1';
注意: This last step can create problems when upgrading via pg_upgrade.

Now you can create a new database:

[postgres]$ createdb blog

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

\l
                              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     |

Enable data checksumming[编辑 | 编辑源代码]

If your database files reside on a file system without checksumming, its data is suspectible to silent data corruption due to bit rot and broken hardware. While those events are rare, you might want to enable PostgreSQL's built-in data checksumming if you care about data integrity. This feature must be enabled on the cluster level, not per-database or per-table.

注意: This feature has a number of caveats:
  • There is a minimal performance impact, especially while reading large datasets from disk. In-memory operations are not affected.
  • PostgreSQL is unable to repair corrupt data - it will only abort transactions reading from corrupt pages to prevent further damage or invalid execution results.
  • Checksums cover on-disk data (row) pages only, not metadata or control structures. In-memory pages are not checksummed. Error-corrected storage and ECC memory is still beneficial.
  • To enable checksumming during cluster creation, add the --data-checksums argument to initdb.
  • To verify whenever checksumming is enabled, run [postgres]$ psql -c "SHOW data_checksums" (which should print off or on).
  • To toggle checksumming on an existing cluster:
  1. Stop postgresql.service.
  2. Run [postgres]$ pg_checksums --pgdata /var/lib/postgres/data --enable (or --disable if you no longer want checksumming). Enabling checksums will rewrite all database pages, which will take a while for large database instances.
  3. Start postgresql.service.

图形化管理工具[编辑 | 编辑源代码]

  • phpPgAdmin — Web-based administration tool for PostgreSQL.
https://github.com/phppgadmin/phppgadmin || phppgadminAUR
  • pgAdmin — Comprehensive design and management GUI for PostgreSQL.
https://www.pgadmin.org/ || pgadmin3AUR or pgadmin4[损坏的链接:package not found]
  • pgModeler — Graphical schema designer for PostgreSQL.
https://pgmodeler.io/ || pgmodelerAUR
  • Postbird — Cross-platform PostgreSQL GUI client, written in JavaScript, runs with Electron.
https://github.com/paxa/postbird || postbird-binAUR

For tools supporting multiple DBMSs, see List of applications/Documents#Database tools.

升级 PostgreSQL[编辑 | 编辑源代码]

这篇文章的某些内容需要扩充。

原因: How to upgrade when using third party extensions? (在 Talk:PostgreSQL#pg_upgrade problem if extensions (like postgis) are used 中讨论)

Upgrading major PostgreSQL versions (e.g. version 14.x to version 15.y) requires some extra maintenance.

注意: Official PostgreSQL upgrade documentation should be followed.
警告: The following instructions could cause data loss. Do not run the commands below blindly, without understanding what they do. Backup database first.

Get the currently used database version via

# cat /var/lib/postgres/data/PG_VERSION

To ensure you do not accidentally upgrade the database to an incompatible version, it is recommended to skip updates to the PostgreSQL packages.

Minor version upgrades are safe to perform. However, if you do an accidental upgrade to a different major version, you might not be able to access any of your data. Always check the PostgreSQL home page 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.

pg_upgrade[编辑 | 编辑源代码]

The pg_upgrade utility attempts to copy over as much compatible data as possible between clusters and upgrading everything else. It is generally the fastest method to upgrade most instances, although it requires access to binaries for both source and target PostgreSQL versions. Read the pg_upgrade(1) man page to understand what actions it performs. For non-trivial instances (e.g. with streaming replication or log-shipping), read the upstream documentation first.

For those wishing to use pg_upgrade, a postgresql-old-upgrade package is available 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. To upgrade from older versions of PostgreSQL there are AUR packages available, e.g. postgresql-12-upgradeAUR. (You must use the pg_upgrade version packaged with the PostgreSQL version you are upgrading to.)

Note that the database cluster 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 cluster must be initialized using the same parameters as the old one.

When you are ready to begin the upgrade:

  1. While the old database cluster is still online, collect the initdb arguments used to create it. Refer to #Initial configuration for more information.
  2. Stop postgresql.service. (Check the unit status to be sure that PostgresSQL was stopped correctly. If it failed, pg_upgrade will fail too.)
  3. Upgrade postgresql, postgresql-libs, and postgresql-old-upgrade.
  4. Rename the old cluster directory, then create a new cluster and temporary working directory:
    # mv /var/lib/postgres/data /var/lib/postgres/olddata
    # mkdir /var/lib/postgres/data /var/lib/postgres/tmp
    # chown postgres:postgres /var/lib/postgres/data /var/lib/postgres/tmp
    [postgres]$ cd /var/lib/postgres/tmp
    
  5. Initialize the new cluster using the same initdb arguments as were used for the old cluster:
    [postgres]$ initdb -D /var/lib/postgres/data --locale=C.UTF-8 --encoding=UTF8 --data-checksums
  6. Upgrade the cluster, replacing PG_VERSION below, with the old PostgreSQL version number (e.g. 14):
    [postgres]$ pg_upgrade -b /opt/pgsql-PG_VERSION/bin -B /usr/bin -d /var/lib/postgres/olddata -D /var/lib/postgres/data
    注意:
    • If necessary, adjust the configuration files of new cluster (e.g. pg_hba.conf and postgresql.conf) to match the old cluster.
    • If pg_upgrade fails with The source cluster was not shut down cleanly, PostgreSQL was not stopped before attempting the upgrade. Stop it, then restart the cluster with old database binaries to recover old cluster files:
      [postgres]$ /opt/pgsql-PG_VERSION/bin/pg_ctl start -D /var/lib/postgres/olddata && /opt/pgsql-PG_VERSION/bin/pg_ctl stop -D /var/lib/postgres/olddata
      You can now safely retry the pg_upgrade command. If the upgrade process still fails, shut down any database processes, downgrade PostgreSQL packages to their previous versions, recover your previous cluster data from backups and restart the upgrade process.
  7. Start postgresql.service again.
  8. Optional: Run [postgres]$ vacuumdb --all --analyze-in-stages to recalculate query analyzer statistics, which should improve query performance shortly after the upgrade. (Adding --jobs=NUMBER_OF_CPU_CORES argument may improve this command's performance.)
  9. Optional: Back up the /var/lib/postgres/olddata directory in case you need to restore a previous PostgreSQL version.
  10. Delete the /var/lib/postgres/olddata directory with old cluster data.
  11. Delete the /var/lib/postgres/tmp directory.

Caveats[编辑 | 编辑源代码]

本文或本章节的事实准确性存在争议。

原因: This section is not backed up with references and is too informal. See Help:Style#Language register. The intended scope overlaps with #Troubleshooting.(在 Talk:PostgreSQL 中讨论)


pg_upgrade is said to be tricky. On a talk in a recent postgresql conference, the presenter asked everyone that had used pg_upgrade to raise their hands, followed up by the next question - how many have used pg_upgrade and it just worked on the first attempt without any errors? Very few hands to be seen on the second question. This section is intended to be a list of problems people have observed when upgrading PostgreSQL under Archlinux (problems should preferably be reported upstream, too).

  • When upgrading from Postgresql 14 to Postgresql 15, old postgresql did not start - it insisted on fetching data from /var/lib/postgres/data rather than /var/lib/postgres/data.old and threw an error message that the directory was created with pg 15 and not pg 14. Moving the old directory to /var/lib/postgres/data and the new directory to /var/lib/postgres/15 did not help - then things were apparently freezing. What did help was to edit /var/lib/postgres/data.old/postgres.conf and remove data_directory from that file. (Hm ... perhaps a user error to have that thing there at all, if the config file is in the data directory, obviously postgresql has already found the data directory and should not need that directive. Postgresql was installed through puppetlabs-postgresql puppet module, it is managing the configuration).

Manual dump and reload[编辑 | 编辑源代码]

You could also do something like this (after the upgrade and install of postgresql-old-upgrade).

注意:
  • Below are the commands for upgrading from PostgreSQL 14. You can find similar commands in /opt/ for your version of PostgreSQL cluster, provided you have matching version of postgresql-old-upgrade package installed.
  • If you had customized your pg_hba.conf file, you may have to temporarily modify it to allow full access to old database cluster from local system. After upgrade is complete set your customization to new database cluster as well and restart postgresql.service.

Stop postgresql.service

# mv /var/lib/postgres/data /var/lib/postgres/olddata
# mkdir /var/lib/postgres/data
# chown postgres:postgres /var/lib/postgres/data
[postgres]$ initdb -D /var/lib/postgres/data --locale=C.UTF-8 --encoding=UTF8 --data-checksums
[postgres]$ /opt/pgsql-14/bin/pg_ctl -D /var/lib/postgres/olddata/ start
# cp /usr/lib/postgresql/postgis-3.so /opt/pgsql-14/lib/ # Only if postgis installed
[postgres]$ pg_dumpall -h /tmp -f /tmp/old_backup.sql
[postgres]$ /opt/pgsql-14/bin/pg_ctl -D /var/lib/postgres/olddata/ stop

Start postgresql.service

[postgres]$ psql -f /tmp/old_backup.sql postgres

Troubleshooting[编辑 | 编辑源代码]

Improve performance of small transactions[编辑 | 编辑源代码]

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. Beware of the caveats, however.

/var/lib/postgres/data/postgresql.conf
synchronous_commit = off

空闲时防止磁盘写入[编辑 | 编辑源代码]

本文内容或本节内容已经过期。

原因: 已在 PostgreSQL 15 及以上版本弃用。[2] (在Talk:PostgreSQL讨论)

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

/var/lib/postgres/data/postgresql.conf
stats_temp_directory = '/run/postgresql'

pgAdmin 4 issues after upgrade to PostgreSQL 12[编辑 | 编辑源代码]

这一章节正在考虑移除。

原因: pgadmin is no longer packaged. (在 Talk:PostgreSQL 讨论)


If you see errors about string indices must be integers when navigating the tree on the left, or about column rel.relhasoids does not exist when viewing the data, remove the server from the connection list in pgAdmin and add a fresh server instance. pgAdmin will otherwise continue to treat the server as a PostgreSQL 11 server resulting in these issues.

PostgreSQL database unable to start after package update when using extensions[编辑 | 编辑源代码]

The cause in this case is mostly the existing package is not compiled for the newer version (and it may be up-to-date), the solution is rebuilding the package either manually or waiting for an update to the extension package.

Failing to start a PostgreSQL server with the older version of the database while upgrading to the newer version with extensions[编辑 | 编辑源代码]

This is caused because the old version of postgres from the package postgresql-old-upgrade does not have the required extensions (.so files) in its lib directory, the current solution is dirty, and might cause a lot of problems so keep a backup of the database just in case, basically copy the required extension .so files from /usr/lib/postgresql/ to /opt/pgsql-XX/lib/ (remember to replace XX with the major version of postgresql-old-upgrade).

For example, for timescaledb

# cp /usr/lib/postgresql/timescaledb*.so /opt/pgsql-13/lib/
警告: while copying the .so files was enough for me, it might be required to copy more files to the correct directories under /opt/pgsql-XX/

to know the exact files to copy, check the contant of the package of the extension using :

$ pacman -Ql package_name
警告: This is a very dirty solution that may break or cause data loss in the database, so keep a backup