8.2. Pgpool-II + Watchdog Setup Example

8.2.6.1. Clustering mode

Pgpool-II has several clustering modes. To set the clustering mode, backend_clustering_mode can be used. In this configuration example, streaming replication mode is used.

backend_clustering_mode = 'streaming_replication'
   

8.2.6.2. listen_addresses

To allow Pgpool-II and PCP to accept all incoming connections, we set listen_addresses = '*'.

listen_addresses = '*'
pcp_listen_addresses = '*'
   

8.2.6.3. port

Specify the port number Pgpool-II listen on.

port = 9999
   

8.2.6.4. Streaming Replication Check

Specify replication delay check user and password in sr_check_user and sr_check_password. In this example, we leave sr_check_password empty, and create the entry in pool_passwd. See Section 8.2.6.9 for how to create the entry in pool_passwd. From Pgpool-II 4.0, if these parameters are left blank, Pgpool-II will first try to get the password for that specific user from pool_passwd file before using the empty password.

sr_check_user = 'pgpool'
sr_check_password = ''
   

8.2.6.5. Health Check

Enable health check so that Pgpool-II performs failover. Also, if the network is unstable, the health check fails even though the backend is running properly, failover or degenerate operation may occur. In order to prevent such incorrect detection of health check, we set health_check_max_retries = 3. Specify health_check_user and health_check_password in the same way like sr_check_user and sr_check_password.

health_check_period = 5
health_check_timeout = 30
health_check_user = 'pgpool'
health_check_password = ''
health_check_max_retries = 3
   

8.2.6.6. Backend Settings

Specify the PostgreSQL backend information. Multiple backends can be specified by adding a number at the end of the parameter name.

# - Backend Connection Settings -

backend_hostname0 = 'server1'
backend_port0 = 5432
backend_weight0 = 1
backend_data_directory0 = '/var/lib/pgsql/14/data'
backend_flag0 = 'ALLOW_TO_FAILOVER'

backend_hostname1 = 'server2'
backend_port1 = 5432
backend_weight1 = 1
backend_data_directory1 = '/var/lib/pgsql/14/data'
backend_flag1 = 'ALLOW_TO_FAILOVER'

backend_hostname2 = 'server3'
backend_port2 = 5432
backend_weight2 = 1
backend_data_directory2 = '/var/lib/pgsql/14/data'
backend_flag2 = 'ALLOW_TO_FAILOVER'
   

To show "replication_state" and "replication_sync_state" column in SHOW POOL NODES command result, backend_application_name parameter is required. Here we specify each backend's hostname in these parameters. (Pgpool-II 4.1 or later)

...
backend_application_name0 = 'server1'
...
backend_application_name1 = 'server2'
...
backend_application_name2 = 'server3'
   

8.2.6.7. Failover configuration

Specify failover.sh script to be executed after failover in failover_command parameter. If we use 3 PostgreSQL servers, we need to specify follow_primary_command to run after failover on the primary node failover. In case of two PostgreSQL servers, follow_primary_command setting is not necessary.

Pgpool-II replaces the following special characters with the backend specific information while executing the scripts. See failover_command for more details about each character.

failover_command = '/etc/pgpool-II/failover.sh %d %h %p %D %m %H %M %P %r %R %N %S'
follow_primary_command = '/etc/pgpool-II/follow_primary.sh %d %h %p %D %m %H %M %P %r %R'
   

Sample scripts failover.sh and follow_primary.sh are installed in /etc/pgpool-II/. Create failover scripts using these sample files.

[all servers]# cp -p /etc/pgpool-II/failover.sh{.sample,}
[all servers]# cp -p /etc/pgpool-II/follow_primary.sh{.sample,}
[all servers]# chown postgres:postgres /etc/pgpool-II/{failover.sh,follow_primary.sh}
   

Basically, it should work if you change PGHOME according to PostgreSQL installation directory.

[all servers]# vi /etc/pgpool-II/failover.sh
...
PGHOME=/usr/pgsql-14
...

[all servers]# vi /etc/pgpool-II/follow_primary.sh
...
PGHOME=/usr/pgsql-14
...
   

Since user authentication is required to use the PCP command in follow_primary_command script, we need to specify user name and md5 encrypted password in pcp.conf in format "username:encrypted password".

if pgpool user is specified in PCP_USER in follow_primary.sh,

# cat /etc/pgpool-II/follow_primary.sh
...
PCP_USER=pgpool
...
   

then we use pg_md5 to create the encrypted password entry for pgpool user as below:

[all servers]# echo 'pgpool:'`pg_md5 PCP password` >> /etc/pgpool-II/pcp.conf
   

Since follow_primary.sh script must execute PCP command without entering a password, we need to create .pcppass in the home directory of Pgpool-II startup user (postgres user) on each server.

[all servers]# su - postgres
[all servers]$ echo 'localhost:9898:pgpool:<pgpool user password>' > ~/.pcppass
[all servers]$ chmod 600 ~/.pcppass
   

8.2.6.8. Pgpool-II Online Recovery Configurations

Next, in order to perform online recovery with Pgpool-II we specify the PostgreSQL user name and online recovery command recovery_1st_stage. Because Superuser privilege in PostgreSQL is required for performing online recovery, we specify postgres user in recovery_user. Then, we create recovery_1st_stage and pgpool_remote_start in database cluster directory of PostgreSQL primary server (server1), and add execute permission.

recovery_user = 'postgres'
recovery_password = ''
recovery_1st_stage_command = 'recovery_1st_stage'
   

Online recovery sample scriptsrecovery_1st_stage and pgpool_remote_start are installed in /etc/pgpool-II/. Copy these files to the data directory of the primary server (server1).

[server1]# cp -p /etc/pgpool-II/recovery_1st_stage.sample /var/lib/pgsql/14/data/recovery_1st_stage
[server1]# cp -p /etc/pgpool-II/pgpool_remote_start.sample /var/lib/pgsql/14/data/pgpool_remote_start
[server1]# chown postgres:postgres /var/lib/pgsql/14/data/{recovery_1st_stage,pgpool_remote_start}
   

Basically, it should work if you change PGHOME according to PostgreSQL installation directory.

[server1]# vi /var/lib/pgsql/14/data/recovery_1st_stage
...
PGHOME=/usr/pgsql-14
...

[server1]# vi /var/lib/pgsql/14/data/pgpool_remote_start
...
PGHOME=/usr/pgsql-14
...
   

In order to use the online recovery functionality, the functions of pgpool_recovery, pgpool_remote_start, pgpool_switch_xlog are required, so we need to install pgpool_recovery on template1 of PostgreSQL server server1.

[server1]# su - postgres
[server1]$ psql template1 -c "CREATE EXTENSION pgpool_recovery"
   

8.2.6.9. Client Authentication Configuration

Because in the section Before Starting, we already set PostgreSQL authentication method to scram-sha-256, it is necessary to set a client authentication by Pgpool-II to connect to backend nodes. When installing with RPM, the Pgpool-II configuration file pool_hba.conf is in /etc/pgpool-II. By default, pool_hba authentication is disabled, set enable_pool_hba = on to enable it.

enable_pool_hba = on
   

The format of pool_hba.conf file follows very closely PostgreSQL's pg_hba.conf format. Set pgpool and postgres user's authentication method to scram-sha-256.

host    all         pgpool           0.0.0.0/0          scram-sha-256
host    all         postgres         0.0.0.0/0          scram-sha-256
   

The default password file name for authentication is pool_passwd. To use scram-sha-256 authentication, the decryption key to decrypt the passwords is required. We create the .pgpoolkey file in Pgpool-II start user postgres's (Pgpool-II 4.1 or later) home directory. (Pgpool-II 4.0 or before, by default Pgpool-II is started as root)

[all servers]# su - postgres
[all servers]$ echo 'some string' > ~/.pgpoolkey
[all servers]$ chmod 600 ~/.pgpoolkey
     

Execute command pg_enc -m -k /path/to/.pgpoolkey -u username -p to register user name and AES encrypted password in file pool_passwd. If pool_passwd doesn't exist yet, it will be created in the same directory as pgpool.conf.

[all servers]# su - postgres
[all servers]$ pg_enc -m -k ~/.pgpoolkey -u pgpool -p
db password: [pgpool user's password]
[all servers]$ pg_enc -m -k ~/.pgpoolkey -u postgres -p
db password: [postgres user's password]

# cat /etc/pgpool-II/pool_passwd
pgpool:AESheq2ZMZjynddMWk5sKP/Rw==
postgres:AESHs/pWL5rtXy2IwuzroHfqg==
   

8.2.6.10. Watchdog Configuration

Enable watchdog functionality on server1, server2, server3.

use_watchdog = on
   

Specify virtual IP address that accepts connections from clients on server1, server2, server3. Ensure that the IP address set to virtual IP isn't used yet.

delegate_IP = '192.168.137.150'
   

To bring up/down the virtual IP and send the ARP requests, we set if_up_cmd, if_down_cmd and arping_cmd. The network interface used in this example is "enp0s8". Since root privilege is required to execute if_up/down_cmd or arping_cmd command, use setuid on these command or allow Pgpool-II startup user, postgres user (Pgpool-II 4.1 or later) to run sudo command without a password.

Here we configure the following parameters to run if_up/down_cmd or arping_cmd with sudo.

if_up_cmd = '/usr/bin/sudo /sbin/ip addr add $_IP_$/24 dev enp0s8 label enp0s8:0'
if_down_cmd = '/usr/bin/sudo /sbin/ip addr del $_IP_$/24 dev enp0s8'
arping_cmd = '/usr/bin/sudo /usr/sbin/arping -U $_IP_$ -w 1 -I enp0s8'
   

Set if_cmd_path and arping_path according to the command path. If if_up/down_cmd or arping_cmd starts with "/", these parameters will be ignored.

if_cmd_path = '/sbin'
arping_path = '/usr/sbin'
   

Specify all Pgpool-II nodes information for configuring watchdog. Specify pgpool_portX using the port number specified in port in Section 8.2.6.3.

hostname0 = 'server1'
wd_port0 = 9000
pgpool_port0 = 9999

hostname1 = 'server2'
wd_port1 = 9000
pgpool_port1 = 9999

hostname2 = 'server3'
wd_port2 = 9000
pgpool_port2 = 9999
   

Specify the method of lifecheck wd_lifecheck_method and the lifecheck interval wd_interval. Here, we use heartbeat method to perform watchdog lifecheck.

wd_lifecheck_method = 'heartbeat'
                                    # Method of watchdog lifecheck ('heartbeat' or 'query' or 'external')
                                    # (change requires restart)
wd_interval = 10
                                    # lifecheck interval (sec) > 0
                                    # (change requires restart)
   

Specify all Pgpool-II nodes information for sending and receiving heartbeat signal.

heartbeat_hostname0 = 'server1'
                                    # Host name or IP address used
                                    # for sending heartbeat signal.
                                    # (change requires restart)
heartbeat_port0 = 9694
                                    # Port number used for receiving/sending heartbeat signal
                                    # Usually this is the same as heartbeat_portX.
                                    # (change requires restart)
heartbeat_device0 = ''
                                    # Name of NIC device (such like 'eth0')
                                    # used for sending/receiving heartbeat
                                    # signal to/from destination 0.
                                    # This works only when this is not empty
                                    # and pgpool has root privilege.
                                    # (change requires restart)

heartbeat_hostname1 = 'server2'
heartbeat_port1 = 9694
heartbeat_device1 = ''
heartbeat_hostname2 = 'server3'
heartbeat_port2 = 9694
heartbeat_device2 = ''
   

If the wd_lifecheck_method is set to heartbeat, specify the time to detect a fault wd_heartbeat_deadtime and the interval to send heartbeat signals wd_heartbeat_deadtime.

wd_heartbeat_keepalive = 2
                                    # Interval time of sending heartbeat signal (sec)
                                    # (change requires restart)
wd_heartbeat_deadtime = 30
                                    # Deadtime interval for heartbeat signal (sec)
                                    # (change requires restart)
   

When Watchdog process is abnormally terminated, the virtual IP may be "up" on both of the old and new active pgpool nodes. To prevent this, configure wd_escalation_command to bring down the virtual IP on other pgpool nodes before bringing up the virtual IP on the new active pgpool node.

wd_escalation_command = '/etc/pgpool-II/escalation.sh'
                                    # Executes this command at escalation on new active pgpool.
                                    # (change requires restart)
    

The sample script escalation.sh is installed in /etc/pgpool-II/.

[all servers]# cp -p /etc/pgpool-II/escalation.sh{.sample,}
[all servers]# chown postgres:postgres /etc/pgpool-II/escalation.sh
    

Basically, it should work if you change the following variables according to your environment. PGPOOL is tha array of the hostname that running Pgpool-II. VIP is the virtual IP address that you set as delegate_IP. DEVICE is the network interface for the virtual IP.

[all servers]# vi /etc/pgpool-II/escalation.sh
...
PGPOOLS=(server1 server2 server3)
VIP=192.168.137.150
DEVICE=enp0s8
...
    

8.2.6.11. Logging

Since Pgpool-II 4.2, the logging collector process has been implemented. In the example, we enable logging collector.

log_destination = 'stderr'
logging_collector = on
log_directory = '/var/log/pgpool_log'
log_filename = 'pgpool-%Y-%m-%d_%H%M%S.log'
log_truncate_on_rotation = on
log_rotation_age = 1d
log_rotation_size = 10MB
   

Create the log directory on all servers.

[all servers]# mkdir /var/log/pgpool_log/
[all servers]# chown postgres:postgres /var/log/pgpool_log/
   

The configuration of pgpool.conf on server1 is completed. Copy the pgpool.conf to other Pgpool-II nodes (server2 and server3).

[server1]# scp -p /etc/pgpool-II/pgpool.conf root@server2:/etc/pgpool-II/pgpool.conf
[server1]# scp -p /etc/pgpool-II/pgpool.conf root@server3:/etc/pgpool-II/pgpool.conf
  

8.2.8.1. Set up PostgreSQL standby server

First, we should set up PostgreSQL standby server by using Pgpool-II online recovery functionality. Ensure that recovery_1st_stage and pgpool_remote_start scripts used by pcp_recovery_node command are in database cluster directory of PostgreSQL primary server (server1).

# pcp_recovery_node -h 192.168.137.150 -p 9898 -U pgpool -n 1
Password:
pcp_recovery_node -- Command Successful

# pcp_recovery_node -h 192.168.137.150 -p 9898 -U pgpool -n 2
Password:
pcp_recovery_node -- Command Successful
   

After executing pcp_recovery_node command, verify that server2 and server3 are started as PostgreSQL standby server.

# psql -h 192.168.137.150 -p 9999 -U pgpool postgres -c "show pool_nodes"
Password for user pgpool
 node_id | hostname | port | status | pg_status | lb_weight |  role   | pg_role | select_cnt | load_balance_node | replication_delay | replication_state | replication_sync_state | last_status_change
---------+----------+------+--------+-----------+-----------+---------+---------+------------+-------------------+-------------------+-------------------+------------------------+---------------------
 0       | server1  | 5432 | up     | up        | 0.333333  | primary | primary | 0          | false             | 0                 |                   |                        | 2021-10-19 07:00:57
 1       | server2  | 5432 | up     | up        | 0.333333  | standby | standby | 0          | true              | 0                 | streaming         | async                  | 2021-10-19 07:00:57
 2       | server3  | 5432 | up     | up        | 0.333333  | standby | standby | 0          | false             | 0                 | streaming         | async                  | 2021-10-19 07:00:57
(3 rows)
   

8.2.8.2. Switching active/standby watchdog

Confirm the watchdog status by using pcp_watchdog_info. The Pgpool-II server which is started first run as LEADER.

# pcp_watchdog_info -h 192.168.137.150 -p 9898 -U pgpool
Password:
3 3 YES server1:9999 Linux server1 server1

server1:9999 Linux server1 server1 9999 9000 4 LEADER 0 MEMBER #The Pgpool-II server started first became "LEADER".
server2:9999 Linux server2 server2 9999 9000 7 STANDBY 0 MEMBER #run as standby
server3:9999 Linux server3 server3 9999 9000 7 STANDBY 0 MEMBER #run as standby
   

Stop active server server1, then server2 or server3 will be promoted to active server. To stop server1, we can stop Pgpool-II service or shutdown the whole system. Here, we stop Pgpool-II service.

[server1]# systemctl stop pgpool.service

# pcp_watchdog_info -p 9898 -h 192.168.137.150 -U pgpool
Password:
3 3 YES server2:9999 Linux server2 server2

server2:9999 Linux server2 server2 9999 9000 4 LEADER 0 MEMBER    #server2 is promoted to LEADER
server1:9999 Linux server1 server1 9999 9000 10 SHUTDOWN 0 MEMBER #server1 is stopped
server3:9999 Linux server3 server3 9999 9000 7 STANDBY 0 MEMBER   #server3 runs as STANDBY
   

Start Pgpool-II (server1) which we have stopped again, and verify that server1 runs as a standby.

[server1]# systemctl start pgpool.service

[server1]# pcp_watchdog_info -p 9898 -h 192.168.137.150 -U pgpool
Password: 
3 3 YES server2:9999 Linux server2 server2

server2:9999 Linux server2 server2 9999 9000 4 LEADER 0 MEMBER
server1:9999 Linux server1 server1 9999 9000 7 STANDBY 0 MEMBER
server3:9999 Linux server3 server3 9999 9000 7 STANDBY 0 MEMBER
   

8.2.8.3. Failover

First, use psql to connect to PostgreSQL via virtual IP, and verify the backend information.

# psql -h 192.168.137.150 -p 9999 -U pgpool postgres -c "show pool_nodes"
Password for user pgpool:
 node_id | hostname | port | status | pg_status | lb_weight |  role   | pg_role | select_cnt | load_balance_node | replication_delay | replication_state | replication_sync_state | last_status_change
---------+----------+------+--------+-----------+-----------+---------+---------+------------+-------------------+-------------------+-------------------+------------------------+---------------------
 0       | server1  | 5432 | up     | up        | 0.333333  | primary | primary | 0          | false             | 0                 |                   |                        | 2021-10-19 07:08:14
 1       | server2  | 5432 | up     | up        | 0.333333  | standby | standby | 0          | false             | 0                 | streaming         | async                  | 2021-10-19 07:08:14
 2       | server3  | 5432 | up     | up        | 0.333333  | standby | standby | 0          | true              | 0                 | streaming         | async                  | 2021-10-19 07:08:14
(3 rows)
   

Next, stop primary PostgreSQL server server1, and verify automatic failover.

[server1]$ pg_ctl -D /var/lib/pgsql/13/data -m immediate stop
   

After stopping PostgreSQL on server1, failover occurs and PostgreSQL on server2 becomes new primary DB.

# psql -h 192.168.137.150 -p 9999 -U pgpool postgres -c "show pool_nodes"
Password for user pgpool:
 node_id | hostname | port | status | pg_status | lb_weight |  role   | pg_role | select_cnt | load_balance_node | replication_delay | replication_state | replication_sync_state | last_status_change
---------+----------+------+--------+-----------+-----------+---------+---------+------------+-------------------+-------------------+-------------------+------------------------+---------------------
 0       | server1  | 5432 | down   | down      | 0.333333  | standby | unknown | 0          | false             | 0                 |                   |                        | 2021-10-19 07:10:01
 1       | server2  | 5432 | up     | up        | 0.333333  | primary | primary | 0          | false             | 0                 |                   |                        | 2021-10-19 07:10:01
 2       | server3  | 5432 | up     | up        | 0.333333  | standby | standby | 0          | true              | 0                 | streaming         | async                  | 2021-10-19 07:10:03
(3 rows)
   

server3 is running as standby of new primary server2.

[server3]# psql -h server3 -p 5432 -U pgpool postgres -c "select pg_is_in_recovery()"
pg_is_in_recovery 
-------------------
t

[server2]# psql -h server2 -p 5432 -U pgpool postgres -c "select pg_is_in_recovery()"
pg_is_in_recovery 
-------------------
f

[server2]# psql -h server2 -p 5432 -U pgpool postgres -c "select * from pg_stat_replication" -x
-[ RECORD 1 ]----+------------------------------
pid              | 7198
usesysid         | 16385
usename          | repl
application_name | server3
client_addr      | 192.168.137.103
client_hostname  |
client_port      | 40916
backend_start    | 2021-10-19 07:10:03.067241+00
backend_xmin     |
state            | streaming
sent_lsn         | 0/12000260
write_lsn        | 0/12000260
flush_lsn        | 0/12000260
replay_lsn       | 0/12000260
write_lag        |
flush_lag        |
replay_lag       |
sync_priority    | 0
sync_state       | async
reply_time       | 2021-10-19 07:11:53.886477+00
   

8.2.8.4. Online Recovery

Here, we use Pgpool-II online recovery functionality to restore server1 (old primary server) as a standby. Before restoring the old primary server, please ensure that recovery_1st_stage and pgpool_remote_start scripts exist in database cluster directory of current primary server server2.

# pcp_recovery_node -h 192.168.137.150 -p 9898 -U pgpool -n 0
Password: 
pcp_recovery_node -- Command Successful
   

Then verify that server1 is started as a standby.

# psql -h 192.168.137.150 -p 9999 -U pgpool postgres -c "show pool_nodes"
Password for user pgpool:
node_id | hostname | port | status | lb_weight |  role   | select_cnt | load_balance_node | replication_delay | replication_state | replication_sync_state | last_status_change  
---------+----------+------+--------+-----------+-----------+---------+---------+------------+-------------------+-------------------+-------------------+------------------------+---------------------
 0       | server1  | 5432 | up     | up        | 0.333333  | standby | standby | 0          | true              | 0                 | streaming         | async                  | 2021-10-19 07:14:06
 1       | server2  | 5432 | up     | up        | 0.333333  | primary | primary | 0          | false             | 0                 |                   |                        | 2021-10-19 07:10:01
 2       | server3  | 5432 | up     | up        | 0.333333  | standby | standby | 0          | false             | 0                 | streaming         | async                  | 2021-10-19 07:10:03
(3 rows)