PostgreSQL Backup & Restore

If you are using PostgreSQL in a production environment, it is important to take precautions to ensure that your users’ data is not lost. By frequently backing up your database, and/or automating backups with a cron task, you will be able to quickly restore your system in the event that your database is lost or corrupted. Fortunately, PostgreSQL includes tools to make this task simple and easy to manage.

Before You Begin 

You should have a working installation of PostgreSQL on your system before beginning this guide. Go through our How to Install PostgreSQL on Ubuntu guide to install PostgreSQL and create a sample database.

Note

The steps in this guide require root privileges. Be sure to run the steps below as root or with the sudo prefix. For more information on privileges, see our Users and Groups guide.

One-Time SQL Dump 

Single Database 

PostgreSQL provides the pg_dump utility to simplify backing up a single database. This command must be run as a user with read permissions to the database you intend to back up.

1. Log in as the postgres user:

   su - postgres
   

2. Dump the contents of a database to a file by running the following command. Replace dbnamewith the name of the database to be backed up.

   pg_dump dbname > dbname.bak
   

   The resulting backup file, dbname.bak, can be transferred to another host with scp or stored locally for later use.

3. To demonstrate restoring lost data, delete your example database and create an empty database in its place:

   dropdb dbname
   createdb dbname
   

4. Restore the database using psql:

   psql test < dbname.bak
   

   There are several options for the backup format:

   • *.bak: compressed binary format
   • *.sql: plaintext dump
   • *.tar: tarball

   Note

   By default, PostgreSQL will ignore any errors that occur during the backup process. This can result in an incomplete backup. To prevent this, you can run the pg_dump command with the -1 option. This will treat the entire backup procedure as a single transaction, which will prevent partial backups in the event of an error:

   pg_dump dbname > dbname.bak
   

Remote Database

Just as psql allows you to connect to a remote host, pg_dump can be run from a client computer to back up data on a remote server. Use the -h flag to specify the IP address of your  2ND and -pto identify the port on which PostgreSQL is listening:

pg_dump -h 198.51.100.0 -p 5432 dbname > dbname.bak

All Databases

Because pg_dump only creates a backup of one database at a time, it does not store information about database roles or other cluster-wide configuration. To store this information, and back up all of your databases simultaneously, you can use pg_dumpall.

1. Create a backup file:

   pg_dumpall > pg_backup.bak
   

2. Restore all databases from the backup:

   psql -f pg_backup.bak postgres
   

Automate Backups with a Cron Task

You may want to set up a cron job so that your database will be backed up automatically at regular intervals. The steps in this section will set up a cron task that will run pg_dump once every week.

1. Make sure you are logged in as the postgres user:

   su - postgres
   

2. Create a directory to store the automatic backups:

   mkdir -p ~/postgres/backups
   

3. Edit the crontab to create the new cron task:

   crontab -e
   

4. Add the following line to the end of the crontab:

   crontab

   0 * * * 0 pg_dump -U postgres dbname > ~/postgres/backups/dbname.bak

5. Save and exit from the editor. Your database will be backed up at midnight every Sunday. To change the time or frequency of the updates, see our Schedule Tasks with Cron guide.
6. Sample crontjob sheduler