How to Change the PostgreSQL Data Directory on Ubuntu
PostgreSQL
18.12.2024
Reading time: 4 min
Share
Databases tend to grow beyond their original filesystem over time. If they share the same partition as the operating system, it could potentially lead to I/O conflicts.
Devices like network block storage RAID provide redundancy and enhance scalability. Regardless of your goals—whether increasing space or optimizing performance—this guide will assist you in moving the PostgreSQL data directory.
Prerequisites
To follow this guide, you will need:
-
An Ubuntu 22.04 VPS server with a non-privileged user account and
sudoprivileges. -
PostgreSQL installed on your server.
In this guide, we will move the data to a block storage device mounted at /mnt/volume_01. The method described here is universal and will help relocate the data directory to another location in any basic storage.
Step 1. Move the PostgreSQL Data Directory
First, check the current location of the data directory by starting an interactive PostgreSQL session. Here, psql is the command to enter the interactive monitor, and -u postgres instructs sudo to run psql as the system user postgres:
sudo -u postgres psqlIn the PostgreSQL command line, enter this command to show the current directory:
SHOW data_directory;By default, the directory is /var/lib/postgresql/xx/main, where xx is your PostgreSQL version.
Exit the PostgreSQL prompt by typing \q and pressing ENTER.
Before making any changes to the directory, stop PostgreSQL to avoid compromising data integrity:
sudo systemctl stop postgresqlYou won’t be able to check the service status directly through systemctl after stopping it. To ensure the service has been stopped, run the following command:
sudo systemctl status postgresqlThe last line of the output will confirm that PostgreSQL has indeed stopped.
To copy the directory to the new location, use the rsync command. You can add flags: -a preserves the permissions and other attributes of the directory, while -v ensures detailed output so you can track progress. To replicate the original directory structure in the new location, run rsync from the postgresql directory. Creating this postgresql directory at the mount point and preserving ownership by the PostgreSQL user will prevent permission issues during future updates.
Note: If tab completion is enabled, make sure the directory doesn't have a trailing slash. Otherwise,
rsyncwill only copy the directory's contents to the mount point, not the directory itself.
Strictly speaking, the versioned directory (e.g., 16) is unnecessary since the location is explicitly defined in the postgresql.conf file. However, it is recommended to follow the project's conventions, especially if you later need to run multiple versions of PostgreSQL:
sudo rsync -av /var/lib/postgresql /mnt/volume_01Once the copy is complete, rename the original folder with a .bak extension, and don't delete it until the move is complete. This ensures that nothing gets mixed up due to directories with the same name:
sudo mv /var/lib/postgresql/16/main /var/lib/postgresql/10/main.bakNow, we can configure PostgreSQL to access the data directory in the new location.
Step 2. Point to the New Location of the Directory
By default, the value for data_directory is set to /var/lib/postgresql/16/main in the file /etc/postgresql/16/main/postgresql.conf. You need to edit this file to point to the new directory:
sudo nano /etc/postgresql/16/main/postgresql.confNow, locate the line starting with data_directory and change the path to point to the new location. The updated directive will look something like this:
# /etc/postgresql/16/main/postgresql.conf
...
data_directory = '/mnt/volume_01/postgresql/10/main'
...Save the file and close it by pressing CTRL+X, then Y, and finally ENTER. No further configuration is needed for PostgreSQL in the new directory. The only thing left to do at this point is restart the PostgreSQL service and verify that it correctly points to the new data directory.
Step 3. Restart PostgreSQL
After changing the data_directory directive in the postgresql.conf file, start the PostgreSQL server using systemctl:
sudo systemctl start postgresqlCheck the server status:
sudo systemctl status postgresqlIf the service started correctly, you will see a line like this at the end of the output.
Finally, to ensure that the new directory is being used, open the PostgreSQL command line:
sudo -u postgres psqlCheck the value of the data directory again:
SHOW data_directory;The output will confirm that PostgreSQL is using the new data directory location. After verifying that everything is working properly, ensure you have access to your database and can interact with its data without issues. Once you are confident that the existing data is intact, you can delete the backup directory:
sudo rm -Rf /var/lib/postgresql/16/main.bakThus, you have successfully moved the PostgreSQL data directory to a new location.
Conclusion
If you followed the instructions correctly, your database directory is now in a new location, and you're closer to being able to scale your storage. Congrats!
PostgreSQL
18.12.2024
Reading time: 4 min
Share
Similar
PostgreSQL
How to Migrate a PostgreSQL Database to Another Server
When working with PostgreSQL, you might eventually need to transfer a database from one server to another. Despite seeming complex, PostgreSQL migration can be accomplished using PostgreSQL's built-in tools.
This article outlines various methods to transfer a PostgreSQL database from one server to another on Ubuntu 22.04.
Prerequisites
To migrate a Postgres database, you will need:
A current server running Ubuntu 22.04 with PostgreSQL installed. The database to be transferred should already exist in PostgreSQL.
A new cloud server or virtual machine with Ubuntu 22.04 and affordable cloud PostgreSQL pre-installed. This article uses PostgreSQL version 15. On Hostman, you can easily deploy such a server by choosing the PostgreSQL image when creating a server.
Optionally, the pgAdmin client application for connecting and managing PostgreSQL databases. PgAdmin can be installed on any OS, including Windows, Linux, and macOS.
Preparation Before Database Transfer
Before starting the Postgres migration, ensure that both servers have:
A network connection between them. Use tools like ping and telnet to verify this (with telnet, check ports 22 and 5432).
Ports 22 (SSH) and 5432 (PostgreSQL) open.
Enough free disk space.
Configuring PostgreSQL for Remote Connections
Ensure PostgreSQL can accept remote connections on both servers:
Edit the postgresql.conf file. If using a different version, replace 15 with your version number:
nano /etc/postgresql/15/main/postgresql.conf
Find the listen_addresses parameter. By default, it is commented out and accepts only local connections (localhost). Allow remote connections from specific addresses or all addresses (for testing purposes, use *):
listen_addresses = '*'
In production environments, specify only the required addresses.
Save and exit the file.
Edit the pg_hba.conf file:
nano /etc/postgresql/15/main/pg_hba.conf
Find the line for IPv4 local connections (# IPv4 local connections) and update it to allow connections from specific addresses or all addresses for testing:
host all all 0.0.0.0/0 md5
Save and exit the file.
Restart the PostgreSQL server:
systemctl restart postgresql
Set a strong password for the PostgreSQL user on both servers:
sudo -i -u postgres psql
In the psql console run:
ALTER USER postgres WITH PASSWORD 'nsH7z*0kl>&7?7';
Where postgres is the username and nsH7z*0kl>&7?7 is the password.
Transferring the Database Using pg_dump and psql
Typically, transferring a database involves three steps:
Creating a backup of the database.
Transferring the backup to the target server.
Restoring the database on the new server.
These steps can be performed using PostgreSQL's pg_dump and psql tools.
Transferring a Database
For instance, let's look at transferring a database named e_commerce from a server with IP 166.1.227.252 to a server with IP 91.206.179.207.
Create the target database beforehand:
CREATE DATABASE e_commerce;
Then run:
pg_dump -h localhost -U postgres e_commerce | psql -h 91.206.179.207 -U postgres e_commerce
Explanation:
pg_dump creates a backup of the database.
-h localhost specifies the database server address.
-U postgres specifies the username.
e_commerce is the database name on the current and new server.
psql connects to the remote PostgreSQL server and loads the database.
-h 91.206.179.207 specifies the target server address.
Transferring Without Remote Access
If remote access is unavailable, save the database to a file, transfer it using scp, and restore it on the target server:
pg_dump -h localhost -U postgres e_commerce > e_commerce.sql && scp e_commerce.sql db-admin@91.206.179.207:/var/lib/postgresql
When executing the command, the password for the postgres system user will be requested, not the password for the postgres user defined within the database.
Where:
pg_dump creates a database backup;
-h localhost is the address of the server (IP address or domain name) where the database is located. In this example, the database server address matches the server address itself. If the database is on another server and network access is available, you can specify the address of the remote database;
-U postgres is the username used to connect to the database;
e_commerce is the name of the database to be transferred;
e_commerce.sql is the name of the file in .sql format where the database will be saved;
scp is a utility for secure file copying between hosts. It uses the SSH protocol for data transfer and protection;
db-admin@91.206.179.207:/var/lib/postgresql means username_on_remote_server@address_of_remote_server:full_path where the backup file will be saved.
After entering the command, you first need to enter the password for the database user account (in this example, it is the postgres user), and then enter the password for the remote server user (in this example, it is the db-admin user).
Now you need to upload the file to the database. Run these commands on the target server.
Create a database in psql:
CREATE DATABASE e_commerce;
Then, exit psql and run in the terminal:
psql e_commerce < e_commerce.sql
Creating a Compressed Archive
For larger databases, create a compressed archive:
pg_dump -h localhost -U postgres e_commerce > e_commerce.tar.gzip && scp e_commerce.tar.gzip db-admin@91.206.179.207:/var/lib/postgresql
Restore from the archive:
psql e_commerce < e_commerce.tar.gzip
Adding a Timestamp to the Archive Name
If you need to know the exact date and time that the database was backed up, you can include it in the name of the file being created.
To do this, use the date command and the date format. The example below will use the day-month-year date format:
pg_dump -h localhost -U postgres e_commerce > e_commerce_$(date +%d-%m-%y).sql
Transferring the Database Using pgAdmin
Alternatively, you can use pgAdmin's graphical interface for the Postgres database migration.
Backup Database
Launch pgAdmin: Open pgAdmin and connect to your PostgreSQL server.
Register Server: Right-click on Object Explorer, select Register, then Server.
Configure Connection:
Name: In the General tab, enter a name for the connection (e.g., my_db).
Next, go to the Connection tab and specify:
Host name/address: Specify the IP address or domain name of the PostgreSQL server.
Port: Default is 5432; change if needed.
Maintenance database: Name of the database for backup.
Username and Password: Enter credentials for database access.
Connect: Click Save to connect. If successful, the database appears on the left sidebar.
Backup Database:
Right-click on the database name and select Backup.
Set a Filename for the backup file.
Choose a Format and Encoding (UTF8 recommended).
Select specific Objects to include.
Click Backup to start.
Restore Database
Prepare New Database:
Open psql shell.
Execute:
CREATE DATABASE e_commerce;
Connect to PostgreSQL Server:
In pgAdmin, connect to the new PostgreSQL server, selecting e_commerce as the database.
Restore Database:
Right-click on the database name and choose Restore.
Set the Format (ensure it matches the backup file).
Specify the Filename of the backup file.
Click Restore to begin.
Wait for the Process completed confirmation.
Conclusion
PostgreSQL offers several methods to migrate databases between servers, including using built-in tools for flexible and robust database backups and transfers.
If you've ran into some troubles, check our instruction on how to create server on Ubuntu.
15 May 2025 · 6 min to read
PostgreSQL
Installing PostgreSQL on Debian
PostgreSQL is an advanced open-source relational database management system (DBMS). There are two ways to install it on Debian 11: from the operating system repository or the official PostgreSQL repository.
In this article, you will learn both methods, as well as how to perform common operations like creating roles and databases. Don't forget to check how to configure static IP address on Debian.
Installation from the Debian 11 repository
On Debian, you can install PostgreSQL directly from the system repository.
First, update your package list. Launch the terminal and run:
sudo apt update && sudo apt upgrade
The PostgreSQL package is available in the Debian repository, so you can install it using the apt utility. To do this, run:
sudo apt install postgresql postgresql-contrib
Once the installation is complete, check the status of the service using the command:
sudo systemctl status postgresql
If the service does not start automatically, you can start it manually. To do this, run:
sudo systemctl start postgresql
To stop a running service, run:
sudo systemctl stop postgresql
Before configuring PostgreSQL on Debian, make sure the service is running.
Installation from the official PostgreSQL repository
If you want to use only the latest versions of Postgres, we recommend using the official PostgreSQL repository for installation and subsequent updates.
First of all, you need to add the GPG signing key. This is a security requirement to verify the authenticity of the PostgreSQL repository. To do this, launch a terminal and run:
curl -fsSL https://www.postgresql.org/media/keys/ACCC4CF8.asc | sudo gpg --dearmor -o /usr/share/keyrings/postgresql-keyring.gpg
Now you are ready to add the Postgres repository. Use the following command:
echo "deb [signed-by=/usr/share/keyrings/postgresql-keyring.gpg] http://apt.postgresql.org/pub/repos/apt/bulseye-pgdg main" | sudo tee /etc/apt/sources.list.d/postgresql.list
After successfully integrating the PostgreSQL library, you can install the DBMS. But before you do that, update the system repository using the command:
sudo apt update
After updating, run the following command to install PostgreSQL on Debian:
sudo apt install postgresql
Installation is completed. Now you can proceed to the basic configuration of PostgreSQL.
Basic setup
When installing Postgres, the postgres user is automatically created. You can use this account for your first connection.
Switch to the postgres user:
sudo su - postgres
Run the psql utility which is a shell for managing PostgreSQL:
psql
You can now interact with the PostgreSQL server. To exit the shell, enter:
\q
You can use the following command to access the Postgres command line without switching users:
sudo -u postgres psql
However, the postgres user is usually only used from localhost. If, for example, you use cloud databases, it is better to create a new role for the connection.
Creating a role and a database
The createuser command allows you to create new roles from the command line. Only superusers and roles with CREATEROLE privileges can create new roles.
In the following example, we will create a new role named hostman and a database named hostman_db, and then grant the new role privileges to manage the database.
First create a new role:
sudo su - postgres -c "createuser hostman"
Then create a new database:
sudo su - postgres -c "createdb hostman_db"
To grant the user permissions to the database, connect to the shell:
sudo -u postgres psql
Run the following query to grant the hostman user privileges to manage the hostman_db database:
GRANT ALL PRIVILEGES ON DATABASE hostman_db TO hostman;
You can create new roles and databases in the PostgreSQL shell. In this case, the syntax will be slightly different.
To create a new role with a password, run:
create user cloud with password 'hostmancloud';
To create a new database, run:
create database cloud_db;
Then you must also grant all privileges with the GRANT ALL PRIVILEGES ON DATABASE … TO … command.
Setting up remote access
By default, the Postgres server only listens on the local interface 127.0.0.1. This may be inconvenient. Let's say you have a server on Hostman with PostgreSQL installed. It will be much more convenient to connect to it remotely. To do this, you need to configure the server to listen to other network interfaces.
To change the configuration, open the postgresql.conf file using any editor. This example uses the nano editor:
sudo nano /etc/postgresql/12/main/postgresql.conf
Find the CONNECTIONS AND AUTHENTICATION section and the line #listen_addresses = 'localhost' in the configuration file. Change the line value to listen_addresses = '*'. If you want the server to listen not to all network interfaces, but only to the selected one, specify it instead of an asterisk.
Save the file and restart the Postgres service for the changes to take effect:
sudo service postgresql restart
The last step is to allow connections from the network. To install it, you need to edit the pg_hba.conf file. Open it in the editor:
sudo nano /etc/postgresql/12/main/pg_hba.conf
Find the IPv4 local connections line. Specify the desired network. For example, like this:
TYPE
DATABASE
USER
ADDRESS
METHOD
host
all
hostman
38.62.228.244
md5
You can use other authentication methods. For a complete list, see the PostgreSQL documentation.
Conclusion
There are two ways to install managed PostgreSQL on Debian.
The first option is to use the system repository. Its main advantage is speed. There is no need to install anything additional, just run one command. The downside is that the system repository does not always contain the latest version of the software.
The second installation option is to use the official PostgreSQL repository. This method ensures that you are using the latest version of the DBMS. But you will have to perform a few more steps: first, add the official repository itself and only then install Postgres from it.
09 May 2025 · 5 min to read
PostgreSQL
Upgrading PostgreSQL Version
In PostgreSQL, version upgrading is performed in different ways:
Installing packages with a new version of PostgreSQL. It is suitable only for minor updates. Before performing the update, study the release notes;
Using the standard pg_dumpall program. It is a reliable method, but there may be a long downtime.
Using the standard pg_upgrade program. Another quick way to upgrade, but errors may occur.
Updating via logical replication. This option has minimal downtime but is only suitable for PostgreSQL versions greater than 10.0. Earlier versions require extensions.
The choice depends on which release is used on the server and which version you want to upgrade to.
In this article, we will look at all the above upgrade methods. If you have some troubles while using PostgreSQL, check our instruction on how to set up your database.
Important notes before upgrading
The main thing is to understand the peculiarities of updates between different versions.
The numbers consist of two digits, for example, 10.1. The first digit is the major version number (10). The second digit is the minor release number (1).
Before PostgreSQL 10, the version numbers consisted of three digits. For example, in 9.6.3, 9.6 is the major release number and 3 is the minor version number.
You must understand this difference to choose the right upgrade method.
In minor versions, the data storage format does not change, so there are no compatibility problems. Therefore, the transition from PostgreSQL 10.1 to 10.6 can be carried very easily. To upgrade, turn off the server, replace the executable files, and start the server again.
However, the documentation notes that some releases may require manual changes. Therefore, always read the release notes before upgrading.
In major versions, the data format may change. This makes updating more difficult. You need to either unload the data and upload it again, use the pg_upgrade program, or use logical replication.
We'll talk about all these methods below.
Upgrading within one major version
Let's look at how to update Postgresql Version 14.1 to PostgreSQL 14.3 on an Ubuntu server.
PostgreSQL is available on Ubuntu by default. Start with the command:
sudo apt-get upgrade
Debian and Ubuntu only release one version of PostgreSQL per OS release. For example, Debian Squeeze/6.0 only has PostgreSQL 8.4. If you need a different version of PostgreSQL, use packages from PGDG.
If you want the latest version of PostgreSQL, you must first install the Postgres repository.
Add a repository:
sudo sh -c 'echo "deb http://apt.postgresql.org/pub/repos/apt $(lsb_release -cs)-pgdg main" > /etc/apt/sources.list.d/pgdg.list'
Import the signing key:
wget --quiet -O - https://www.postgresql.org/media/keys/ACCC4CF8.asc | sudo apt-key add -
Update the list of packages in the system:
sudo apt-get update
Install the required Postgres version:
sudo apt-get -y install postgresql-14.3
To view the list of installed DBMSs, run:
dpkg --get-selections | grep postgres
Also see the list of clusters:
pg_lsclusters
Stop PostgreSQL before making changes:
sudo service postgresql stop
When Postgres packages are installed, they create a default cluster for you. You should rename the new Postgres cluster so there are no conflicts with the old cluster name when upgrading.
sudo pg_renamecluster 14.3 main main_pristine
Upgrade the old cluster:
sudo pg_upgradecluster 14.1 main
Start the service:
sudo service postgresql start
Check the list of clusters again and make sure the new one is working:
pg_lsclusters
Get rid of the old cluster:
sudo pg_dropcluster 14.1 main
Upgrading via pg_dumpall
If you need to change the major version, use the pg_dumpall program. The essence of this method is to unload data from one main version and then load it into another.
Before unloading data, make sure that no changes are being made to the database right now. Otherwise, some of the changes may not be included in the final dump.
Get the dump and write it to a file:
pg_dumpall > output_file
Stop the server:
pg_ctl stop
Change the name of the old directory to eliminate name conflicts:
mv /usr/local/pgsql /usr/local/pgsql.old
The old directory can simply be deleted. But it would be wiser to rename it to leave room for restoring data. Instead of /usr/local/pgsql, specify the path to your directory.
Install the new version from source. There is a detailed guide in the documentation.
Form a new cluster:
/usr/local/pgsql/bin/initdb -D /usr/local/pgsql/data
Transfer all changes to the pg_hba.conf and postgresql.conf files.
Start the database server:
/usr/local/pgsql/bin/postgres -D /usr/local/pgsql/data
Restore data from backup:
/usr/local/pgsql/bin/psql -d postgres -f output_file
The disadvantage of this method is that the server will be turned off for a long time. To reduce downtime, you can install the new server in a different directory and then run the old and new servers on different ports. To transfer data, use the command:
pg_dumpall -p 5432 | psql -d postgres -p 5433
Instead of 5432 and 5433, specify the port numbers on which you ran the servers.
Upgrading via pg_upgrade
To avoid unloading and uploading data, use the pg_upgrade program. It helps you update faster. The pg_upgrade program creates system tables anew, taking into account changes in the PostgreSQL latest versions. In this case, the old data files are retained. In addition, pg_upgrade verifies that the clusters are compatible.
The pg_upgrade program helps you upgrade from PostgreSQL 8.4.X to the current DBMS release.
Let's look at a general upgrade plan using pg_upgrade.
Move the old cluster. This must be done if the directory was not linked to the old release. For example, it is located at /usr/local/pgsql. If you do not rename it, a conflict will occur during the update.
Renaming is performed when the server is turned off with the command:
mv /usr/local/pgsql /usr/local/pgsql.old
Build a new version from source. Adjust configure so that the flags are compatible with the old cluster configuration. Before starting the upgrade, pg_upgrade will check the compatibility of the flags.
Install new executable files. To place the server in a non-standard directory, use prefix:
make prefix=/usr/local/pgsql.new install
Use initdb to initialize a new cluster. Check that the flags are compatible with the flags of the old cluster.
Install extension shared object files. If updates are available, pg_upgrade will automatically create a script for their subsequent installation.
Transfer full text search files.
Configure peer authentication in pg_hba.conf. This needs to be done because pg_upgrade will connect to the old and new hosts multiple times.
Check that the old and new servers are stopped:
pg_ctl -D /opt/PostgreSQL/9.6 stoppg_ctl -D /opt/PostgreSQL/14 stop
Run pg_upgrade from the new server. First, call the program in check mode by adding the flag:
pg_upgrade --check
In response, you will receive a list of adjustments that need to be made manually after the upgrade.
If you made changes to the pg_hba.conf file, restore it to its previou state. Sometimes, other configurations on the new cluster need to be changed to match the settings of the old cluster.
Start the server.
If you are satisfied with the result, delete the old cluster.
For details and features of using pg_upgrade, see the documentation.
Using Logical Replication
Starting from the 10th version of PostgreSQL, we have a method for logical replication of data objects and changes in them. It is based on the use of replication identifiers; usually, they are primary keys.
Logical replication uses the publish-subscribe model. The user creates a snapshot of the published database and copies it to the subscriber. In the PostgreSQL documentation, one of the typical scenarios for using logical replication is precisely the replication between different major versions of PostgreSQL.
The backup server can be located on the same or different host. After synchronization is complete, various options are available. For example, you can make the new server the main one and turn off the old one.
The main advantage of logical replication is minimal downtime. Server interruption, in some cases, is limited to a few seconds.
Conclusion
We looked at universal methods for upgrading PostgreSQL to a new major version and updating within one major version.
Using this guide, you can select the method that works best for your particular case and update your current version of PostgreSQL.
Hostman provides pre-configured and ready-to-use cloud databases, including cloud PostgreSQL.
04 May 2025 · 7 min to read
Do you have questions,
comments, or concerns?
Our professionals are available to assist you at any moment,
whether you need help or are just unsure of where to start.
whether you need help or are just unsure of where to start.
