On Debian-based systems like Ubuntu, PostgreSQL uses pg_ctlcluster for cluster management, which is part of the postgresql-common package. The service management is handled through systemd, with the main service file located at /lib/systemd/system/postgresql.service.

First, list all available PostgreSQL clusters:

pg_lsclusters

Output example:

Ver Cluster Port Status Owner    Data directory               Log file
12  main    5432 online postgres /var/lib/postgresql/12/main  /var/log/postgresql/postgresql-12-main.log
13  main    5433 down   postgres /var/lib/postgresql/13/main  /var/log/postgresql/postgresql-13-main.log

To prevent a specific cluster from starting automatically while keeping it installed:

1. Create a systemd override file for the cluster:

sudo systemctl edit postgresql@<version>-<cluster_name>

For example, for PostgreSQL 12's main cluster:

sudo systemctl edit postgresql@12-main

Add these lines to the override file:

[Unit]
RefuseManualStart=yes

[Service]
ExecStart=
ExecStart=/bin/true

Check the service status:

systemctl status postgresql@12-main

The output should show the service as masked or inactive.

You can also manually control cluster startup:

sudo pg_ctlcluster 12 main stop
sudo -u postgres pg_dropcluster --stop 12 main

To temporarily prevent auto-start while keeping the cluster:

sudo pg_ctlcluster 12 main disable

When working with multiple clusters, you can specify which one to start:

sudo pg_ctlcluster 13 main start

To make your new cluster the default (starting on port 5432):

sudo pg_renamecluster 13 main main_new
sudo pg_renamecluster 12 main main_old

Before making these changes:

# Backup the cluster
sudo -u postgres pg_dumpall -c -f /path/to/backup.sql

# Verify the backup
ls -lh /path/to/backup.sql

To reactivate a disabled cluster later:

sudo systemctl unmask postgresql@12-main
sudo systemctl enable postgresql@12-main

---

When working with PostgreSQL on Ubuntu, you might encounter situations where you need to maintain an existing cluster (with potentially incorrect configurations) while creating a new properly configured cluster. A common case is when the original cluster was created with wrong locale settings like en_GB.UTF-8 when you actually need C collation.

First, check your existing clusters:

pg_lsclusters

This will display output similar to:

Ver Cluster Port Status Owner    Data directory               Log file
12  main    5432 online postgres /var/lib/postgresql/12/main  /var/log/postgresql/postgresql-12-main.log

Create a new cluster with correct parameters:

sudo pg_createcluster 12 newcluster --locale=C --start

Verify the new cluster is running:

pg_lsclusters

To prevent the old cluster from auto-starting while keeping it installed:

sudo pg_ctlcluster 12 main stop
sudo systemctl disable postgresql@12-main

The postgresql service is actually a meta-service that starts all enabled clusters. To control individual clusters:

sudo systemctl list-unit-files | grep postgresql

To start/stop specific clusters:

# Start only the new cluster
sudo systemctl start postgresql@12-newcluster

# Start all enabled clusters (excluding our disabled main cluster)
sudo service postgresql start

After making these changes:

# Check cluster status
pg_lsclusters

# Verify systemd configuration
systemctl is-enabled postgresql@12-main

• The old cluster remains fully intact in its data directory
• You can manually start it with sudo systemctl start postgresql@12-main when needed
• All configuration files remain unchanged
• This method preserves the ability to quickly revert if needed

You can also manage clusters directly:

# Disable auto-start at the PostgreSQL level
sudo pg_ctlcluster 12 main disable

# Re-enable when needed
sudo pg_ctlcluster 12 main enable