Configuring PGpool-II is a crucial step in deploying a PostgreSQL database cluster. In this guide, we will walk through the steps to configure PGpool-II on an Ubuntu Linux operating system, helping you maximize the performance and high availability of your database.

Installation Architecture

Before we begin, we need to prepare 4 servers:

Step 3: Configure Load Balancing and Replication

You need to configure Replication first, if you haven’t configured it, refer to Setting Up PostgreSQL 16 Replication.

Add the user replicator:

Create the file pcp.conf:

password_hash=$(pg_md5 PassKhongChilaPasss) && echo "postgre:$password_hash" >> /etc/pgpool2/pcp.conf

Next, configure the file pgpool.conf on the pgpool2 server:

sudo chown pgpool:pgpool /etc/pgpool2/pgpool.conf
sudo chmod 0600 /etc/pgpool2/pgpool.conf

Then, open the file pgpool.conf with a text editor nano:

sudo nano /etc/pgpool2/pgpool.conf

Add the following configuration to the pgpool.conf file:

listen_addresses = '*' 
port = 9999 

backend_hostname0 = '192.168.56.11'
backend_port0 = 5432
backend_weight0 = 0
backend_data_directory0 = '/home/ubuntu/postgresql/master'
backend_flag0 = 'ALLOW_TO_FAILOVER'
backend_application_name0 = 'postgresql-master'

backend_hostname1 = '192.168.56.12'
backend_port1 = 5432
backend_weight1 = 1
backend_data_directory1 = '/home/ubuntu/postgresql/slave-01'
backend_flag1 = 'ALLOW_TO_FAILOVER'
backend_application_name1 = 'postgresql-slave-01'

backend_hostname2 = '192.168.56.13'
backend_port2 = 5432
backend_weight2 = 2
backend_data_directory2 = '/home/ubuntu/postgresql/slave-02'
backend_flag2 = 'ALLOW_TO_FAILOVER'
backend_application_name2 = 'postgresql-slave-02'

log_statement = on
log_per_node_statement = on

sr_check_user = 'replicator'
health_check_user = 'replicator'
health_check_period = 10

# Process management configuration
process_management_mode = static
process_management_strategy = lazy
num_init_children = 320
min_spare_children = 1000
max_spare_children = 5000
max_pool = 1000

In this configuration:

• listen_addresses is the IP address PGpool-II will listen on, port is the port PGpool-II will listen on.
• backend_hostname0, backend_port0, backend_weight0, backend_data_directory0, backend_flag0, backend_application_name0 are the connection details to the postgresql-master server.
• backend_hostname1, backend_port1, backend_weight1, backend_data_directory1, backend_flag1, backend_application_name1 are the connection details to the postgresql-slave-01 server.
• backend_hostname2, backend_port2, backend_weight2, backend_data_directory2, backend_flag2, backend_application_name2 are the connection details to the postgresql-slave-02 server.
• log_statement and log_per_node_statement are the log query configurations.
• sr_check_user and health_check_user are the usernames for database server health checks. replicator is the user we created in the step Setting Up PostgreSQL 16 Replication.
• backend_flag is the flag that allows FAILOVER when the postgresql-master server fails, ALLOW_TO_FAILOVER is the flag that allows FAILOVER when the postgresql-slave-01 and postgresql-slave-02 servers fail. Other flags:
  • DISALLOW_TO_FAILOVER: does not allow FAILOVER
  • ALWAYS_MASTER: always the postgresql-master server
• process_management_mode is the process management mode, static is the static mode, dynamic is the dynamic mode.
  • static: The number of processes is statically managed and does not change.
  • dynamic: The number of processes is dynamically managed depending on the system load.
• process_management_strategy is the process management strategy, lazy is the lazy mode, smart is the smart mode.
  • lazy: Only create processes when needed.
  • smart: Create processes in advance to avoid delays when needed.
• num_init_children is the initial number of processes, if below it will create more processes.
• min_spare_children is the minimum number of processes, if below it will create more processes.
• max_spare_children is the maximum number of processes, if exceeded they will be killed.
• max_pool is the maximum number of connections, if exceeded connections will be rejected.

Step 4: Run PGpool-II Configuration Check

Finally, run the PGpool-II configuration check:

sudo /usr/sbin/pgpool -n -f /etc/pgpool2/pgpool.conf -F /etc/pgpool2/pcp.conf

After configuration, you can check the status of PGpool-II by accessing it through pg4admin:

• ip_address is the IP address (192.168.56.5) of the pgpool2 server.
• username is the default username postgresql.
• password is the password for the postgresql account.

Step 5: Create Test Data

Here we will create a student database and 10000 test data:

# Create the student database
CREATE TABLE student (
    student_id SERIAL PRIMARY KEY,
    first_name VARCHAR(50) NOT NULL,
    last_name VARCHAR(50) NOT NULL,
    email VARCHAR(100),
    date_of_birth DATE,
    grade_level INT CHECK (grade_level BETWEEN 1 AND 12)
);

WITH generated_data AS (
       SELECT 
           substring(md5(random()::text) from 1 for 8) AS first_name,
           substring(md5(random()::text) from 1 for 8) AS last_name,
           concat(substring(md5(random()::text) from 1 for 6), '@example.com') AS email,
           '1990-01-01'::date + random() * (age(now(), '1990-01-01'::date)) AS date_of_birth,
           (random() * 11 + 1)::int AS grade_level
       FROM generate_series(1,1000000) -- Generate 100 rows
   )
   INSERT INTO student (first_name, last_name, email, date_of_birth, grade_level)
   SELECT * FROM generated_data; 

As shown in the image, the INSERT statement has been executed on the postgresql-master server.

Next, we query with the SELECT statement on PG4Admin:

SELECT * FROM student;

Then we check the log of the pgpool2 server:

As shown in the image, the SELECT statement has been executed on the postgresql-slave-01 server through the pgpool2 server.

Thus, we have successfully configured PGpool-II for Load Balancing and Replication.