Перейти к основному содержимому

Задание 2: Включение контроля целостности передачи WAL на реплику

В этом задании мы выполним подготовительные действия, а затем последовательно изучим и включим режим проверки контрольных сумм, режим декодирования и гибридный режим.

Подготовительные действия​

Получим топологию кластера:

[student@srv1 ~]$ sudo su - postgres
===============================================================
 You are connected to Pangolin Cluster (Pobeda)
 - Hostname             : srv1
 - IP Address           : <IP-Address1>
---------------------------------------------------------------
 - node ( master )      : <IP-Address1>
 - node ( replica )     : <IP-Address2>
 - arbiter              : <IP-Address3>
 - Manager version      : 2.1.13
 - Pangolin version     : 6.7.0
===============================================================
[postgres@srv1 ~]$ list
+ Cluster: Pobeda (7576294836334315678) +---------+-----------+----+-----------+
| Member        | Host                  | Role    | State     | TL | Lag in MB |
+---------------+-----------------------+---------+-----------+----+-----------+
| <IP-Address1> | <IP-Address1>:5433    | Leader  | running   | 10 |           |
| <IP-Address2> | <IP-Address2>:5433    | Replica | streaming | 10 |         0 |
+---------------+-----------------------+---------+-----------+----+-----------+
[postgres@srv1 ~]$ exit

В примере видно, что srv1 - мастер, srv2 - реплика.

Проверим статус службы pangolin-manager.service на мастере:

[student@srv1 ~]$ sudo systemctl status pangolin-manager.service
Вывод команды:

● pangolin-manager.service - Runners to orchestrate a high-availability PostgreSQL
     Loaded: loaded (/usr/lib/systemd/system/pangolin-manager.service; enabled; preset: disabled)
    Drop-In: /usr/lib/systemd/system/service.d
             └─10-timeout-abort.conf
     Active: active (running) since  Fri 2026-09-18 16:36:04 MSK; 2 days ago
    Process: 651 ExecStartPre=/bin/mkdir -p /var/run/postgresql (code=exited, status=0/SUCCESS)
    Process: 653 ExecStartPre=/bin/chown -R postgres:postgres /var/run/postgresql (code=exited, status=0/SUCCESS)
   Main PID: 656 (pangolin-manage)
      Tasks: 21 (limit: 4667)
     Memory: 262.7M
        CPU: 7min 34.236s
     CGroup: /system.slice/pangolin-manager.service
             ├─656 /opt/pangolin-manager/bin/pangolin-manager-bin/pangolin-manager.bin /etc/pangolin-manager/postgres.yml
             ├─770 /usr/pangolin-6.7/bin/postgres -D /pgdata/06/data/ --config-file=/pgdata/06/data/postgresql.conf --listen_addresses=0.0.0.0 -->
             ├─773 "postgres: Pobeda: logger "
             ├─774 "postgres: Pobeda: checkpointer "
             ├─775 "postgres: Pobeda: background writer "
             ├─777 "postgres: Pobeda: idle sessions terminator "
             ├─778 "postgres: Pobeda: authproc "
             ├─779 "postgres: Pobeda: password policy cache "
             ├─782 "postgres: Pobeda: integrity check launcher "
             ├─783 "postgres: Pobeda: license checker "
             ├─787 "postgres: Pobeda: patroni postgres 127.0.0.1(34866) idle"
             ├─802 "postgres: Pobeda: walwriter "
             ├─803 "postgres: Pobeda: autovacuum launcher "
             ├─804 "postgres: Pobeda: pg_cron launcher "
             ├─805 "postgres: Pobeda: logical replication launcher "
             └─841 "postgres: Pobeda: walsender patroni <IP-Address2>(34786) streaming 0/14000000"

На реплике:

[student@srv1 ~]$ ssh srv2 sudo systemctl status pangolin-manager.service
Вывод команды:

● pangolin-manager.service - Runners to orchestrate a high-availability PostgreSQL
     Loaded: loaded (/usr/lib/systemd/system/pangolin-manager.service; enabled; preset: disabled)
    Drop-In: /usr/lib/systemd/system/service.d
             └─10-timeout-abort.conf
     Active: active (running) since  Fri 2026-09-18 16:36:15 MSK; 2 days ago
    Process: 643 ExecStartPre=/bin/mkdir -p /var/run/postgresql (code=exited, status=0/SUCCESS)
    Process: 647 ExecStartPre=/bin/chown -R postgres:postgres /var/run/postgresql (code=exited, status=0/SUCCESS)
   Main PID: 649 (pangolin-manage)
      Tasks: 18 (limit: 4667)
     Memory: 281.0M
        CPU: 21min 24.661s
     CGroup: /system.slice/pangolin-manager.service
             ├─649 /opt/pangolin-manager/bin/pangolin-manager-bin/pangolin-manager.bin /etc/pangolin-manager/postgres.yml
             ├─771 /usr/pangolin-6.7/bin/postgres -D /pgdata/06/data/ --config-file=/pgdata/06/data/postgresql.conf --listen_addresses=0.0.0.0 --port=5433 --cluster_name=Pobeda --wal_level=replica --hot_standby=on --max_connections=111 --max_wal_senders=10 --max_prepared_transactions=3 --max_locks_per_transaction=64 --track_commit_timestamp=False --max_replication_slots=10 --max_worker_processes=32 --wal_log_hints=True --authentication_max_workers=16
             ├─774 "postgres: Pobeda: logger "
             ├─775 "postgres: Pobeda: checkpointer "
             ├─776 "postgres: Pobeda: background writer "
             ├─777 "postgres: Pobeda: startup recovering 0000000E0000000000000013"
             ├─778 "postgres: Pobeda: idle sessions terminator "
             ├─779 "postgres: Pobeda: authproc "
             ├─780 "postgres: Pobeda: password policy cache "
             ├─782 "postgres: Pobeda: integrity check launcher "
             ├─783 "postgres: Pobeda: license checker "
             ├─790 "postgres: Pobeda: patroni postgres 127.0.0.1(40988) idle"
             └─804 "postgres: Pobeda: walreceiver streaming 0/14000000"

На мастере и реплике создадим каталог для хранения копий WAL и файлов с контрольными суммами:

[student@srv1 ~]$ for comp in srv{1,2}; do ssh ${comp} sudo -u postgres mkdir /pgdata/06/wal_backup; done

Проверим:

[student@srv1 ~]$ for comp in srv{1,2}; do ssh ${comp} sudo -u postgres ls -ld /pgdata/06/wal_backup; done
drwxr-xr-x 2 postgres postgres 4096 фев 14 21:26 /pgdata/06/wal_backup
drwxr-xr-x 2 postgres postgres 4096 фев 14 21:26 /pgdata/06/wal_backup

Сохраним копии конфигурационных файлов перед их редактированием:

[student@srv1 ~]$ sudo -u postgres cp -v /etc/pangolin-manager/postgres.yml{,.bak}
'/etc/pangolin-manager/postgres.yml' -> '/etc/pangolin-manager/postgres.yml.bak'
[student@srv1 ~]$ ssh srv2 sudo -u postgres cp -v /etc/pangolin-manager/postgres.yml{,.bak}
'/etc/pangolin-manager/postgres.yml' -> '/etc/pangolin-manager/postgres.yml.bak'

Используя любой текстовый редактор, вставьте настройку параметра wal_sender_backup_directory, указывающую на его местоположение. Например, можно вставить эту настройку перед строкой ssl: on. В примере ее позиция определена с помощью grep -n:

[student@srv1 ~]$ sudo -u postgres grep -n 'ssl: on' /etc/pangolin-manager/postgres.yml
226:        ssl: on

Это 226-я строка в данном случае.

Итак, отредактируем на мастере и реплике конфигурационный файл, вставив туда настройку wal_sender_backup_directory: /pgdata/06/wal_backup.

Проверим, что получилось на мастере:

[student@srv1 ~]$ sudo -u postgres grep -B1 -n 'ssl: on' /etc/pangolin-manager/postgres.yml
226-        wal_sender_backup_directory: /pgdata/06/wal_backup
227:        ssl: on

И на реплике:

[student@srv1 ~]$ ssh srv2
Last login: Sat Feb 14 21:40:20 2026 from <IP-Address1>
===============================================================
 You are connected to Pangolin Cluster (Pobeda)
 - Hostname             : srv2
 - IP Address           : <IP-Address2>
---------------------------------------------------------------
 - node ( master )      : <IP-Address1>
 - node ( replica )     : <IP-Address2>
 - arbiter              : <IP-Address3>
 - Manager version      : 2.1.13
 - Pangolin version     : 6.7.0
===============================================================
[student@srv2 ~]$ sudo -u postgres grep -B1 -n 'ssl: on' /etc/pangolin-manager/postgres.yml
226-        wal_sender_backup_directory: /pgdata/06/wal_backup
227:        ssl: on

Не выходя из сеанса на реплике, остановим pangolin-manager:

[student@srv2 ~]$ sudo systemctl stop pangolin-manager
[student@srv2 ~]$ sudo systemctl status pangolin-manager
○ pangolin-manager.service - Runners to orchestrate a high-availability PostgreSQL
     Loaded: loaded (/usr/lib/systemd/system/pangolin-manager.service; enabled; preset: disabled)
    Drop-In: /usr/lib/systemd/system/service.d
             └─10-timeout-abort.conf
     Active: inactive (dead) since Fri 2026-09-18 21:42:13 MSK; 4s ago
   Duration: 2d 5h 5min 57.063s
    Process: 643 ExecStartPre=/bin/mkdir -p /var/run/postgresql (code=exited, status=0/SUCCESS)
    Process: 647 ExecStartPre=/bin/chown -R postgres:postgres /var/run/postgresql (code=exited, status=0/SUCCESS)
    Process: 649 ExecStart=/opt/pangolin-manager/bin/pangolin-manager-bin/pangolin-manager.bin /etc/pangolin-manager/postgres.yml (code=exited, s>
   Main PID: 649 (code=exited, status=0/SUCCESS)
        CPU: 21min 37.793s

Выйдем из сессии на реплике и на мастере рестартуем этот же сервис:

[student@srv1 ~]$ sudo systemctl restart pangolin-manager
[student@srv1 ~]$ sudo systemctl status pangolin-manager
Вывод команды:

● pangolin-manager.service - Runners to orchestrate a high-availability PostgreSQL
     Loaded: loaded (/usr/lib/systemd/system/pangolin-manager.service; enabled; preset: disabled)
    Drop-In: /usr/lib/systemd/system/service.d
             └─10-timeout-abort.conf
     Active: active (running) since Fri 2026-09-18 21:43:33 MSK; 5s ago
    Process: 49586 ExecStartPre=/bin/mkdir -p /var/run/postgresql (code=exited, status=0/SUCCESS)
    Process: 49588 ExecStartPre=/bin/chown -R postgres:postgres /var/run/postgresql (code=exited, status=0/SUCCESS)
   Main PID: 49589 (pangolin-manage)
      Tasks: 20 (limit: 4667)
     Memory: 183.3M
        CPU: 882ms
     CGroup: /system.slice/pangolin-manager.service
             ├─49589 /opt/pangolin-manager/bin/pangolin-manager-bin/pangolin-manager.bin /etc/pangolin-manager/postgres.yml
             ├─49603 /usr/pangolin-6.7/bin/postgres -D /pgdata/06/data/ --config-file=/pgdata/06/data/postgresql.conf --listen_addresses=0.0.0.0 >
             ├─49606 "postgres: Pobeda: logger "
             ├─49607 "postgres: Pobeda: checkpointer "
             ├─49608 "postgres: Pobeda: background writer "
             ├─49610 "postgres: Pobeda: idle sessions terminator "
             ├─49611 "postgres: Pobeda: authproc "
             ├─49612 "postgres: Pobeda: password policy cache "
             ├─49615 "postgres: Pobeda: integrity check launcher "
             ├─49616 "postgres: Pobeda: license checker "
             ├─49619 "postgres: Pobeda: patroni postgres 127.0.0.1(57198) idle"
             ├─49630 "postgres: Pobeda: walwriter "
             ├─49631 "postgres: Pobeda: autovacuum launcher "
             ├─49632 "postgres: Pobeda: pg_cron launcher "
             └─49633 "postgres: Pobeda: logical replication launcher "

Запустим сервис на реплике:

[student@srv1 ~]$ ssh srv2 sudo systemctl start pangolin-manager
[student@srv1 ~]$ ssh srv2 sudo systemctl status pangolin-manager
Вывод команды:

● pangolin-manager.service - Runners to orchestrate a high-availability PostgreSQL
     Loaded: loaded (/usr/lib/systemd/system/pangolin-manager.service; enabled; preset: disabled)
    Drop-In: /usr/lib/systemd/system/service.d
             └─10-timeout-abort.conf
     Active: active (running) since Fri 2026-09-18 21:44:26 MSK; 5s ago
    Process: 81316 ExecStartPre=/bin/mkdir -p /var/run/postgresql (code=exited, status=0/SUCCESS)
    Process: 81317 ExecStartPre=/bin/chown -R postgres:postgres /var/run/postgresql (code=exited, status=0/SUCCESS)
   Main PID: 81318 (pangolin-manage)
      Tasks: 18 (limit: 4667)
     Memory: 183.4M
        CPU: 892ms
     CGroup: /system.slice/pangolin-manager.service
             ├─81318 /opt/pangolin-manager/bin/pangolin-manager-bin/pangolin-manager.bin /etc/pangolin-manager/postgres.yml
             ├─81333 /usr/pangolin-6.7/bin/postgres -D /pgdata/06/data/ --config-file=/pgdata/06/data/postgresql.conf --listen_addresses=0.0.0.0 --port=5433 --cluster_name=Pobeda --wal_level=replica --hot_standby=on --max_connections=111 --max_wal_senders=10 --max_prepared_transactions=3 --max_locks_per_transaction=64 --track_commit_timestamp=False --max_replication_slots=10 --max_worker_processes=32 --wal_log_hints=True --authentication_max_workers=16
             ├─81336 "postgres: Pobeda: logger "
             ├─81337 "postgres: Pobeda: checkpointer "
             ├─81338 "postgres: Pobeda: background writer "
             ├─81339 "postgres: Pobeda: startup recovering 0000000F0000000000000014"
             ├─81340 "postgres: Pobeda: idle sessions terminator "
             ├─81341 "postgres: Pobeda: authproc "
             ├─81342 "postgres: Pobeda: password policy cache "
             ├─81344 "postgres: Pobeda: integrity check launcher "
             ├─81345 "postgres: Pobeda: license checker "
             ├─81348 "postgres: Pobeda: walreceiver streaming 0/14000218"
             └─81351 "postgres: Pobeda: patroni postgres 127.0.0.1(57032) idle"

Подключимся к СУБД и проверим, установлен ли параметр wal_sender_backup_directory:

[student@srv1 ~]$ sudo -iu postgres psql -c 'SHOW wal_sender_backup_directory'
===============================================================
 You are connected to Pangolin Cluster (Pobeda)
 - Hostname             : srv1
 - IP Address           : <IP-Address1>
---------------------------------------------------------------
 - node ( master )      : <IP-Address1>
 - node ( replica )     : <IP-Address2>
 - arbiter              : <IP-Address3>
 - Manager version      : 2.1.13
 - Pangolin version     : 6.7.0
===============================================================

 wal_sender_backup_directory
-----------------------------
 /pgdata/06/wal_backup
(1 row)

Поскольку пока никакой режим проверки целостности WAL не включен, то каталог для сохранения резервных копий сегментов WAL пока пуст:

[student@srv1 ~]$ sudo ls -l /pgdata/06/wal_backup
итого 0

Включение режима проверки контрольных сумм​

Для включения режима проверки контрольных сумм необходимо:

  1. Установить значение параметра wal_sender_check_crc: on.
  2. Указать режим проверки контрольных сумм wal_sender_check_type: block.

Как и для настройки пути к каталогу сохранения резервных копий WAL, сначала включим эти параметры в конфигурационные файлы на мастере и реплике.

Используем текстовый редактор и внесем изменения в конфигурацию мастера и реплики. Пример конфигурационного файла на мастере:

[student@srv1 ~]$ sudo -u postgres grep -B3 -n 'ssl: on' /etc/pangolin-manager/postgres.yml
226-        wal_sender_backup_directory: /pgdata/06/wal_backup
227-        wal_sender_check_crc: on
228-        wal_sender_check_type: block
229:        ssl: on

И на реплике:

[student@srv1 ~]$ ssh srv2
Last login: Sat Feb 14 21:56:50 2026 from <IP-Address1>
===============================================================
 You are connected to Pangolin Cluster (Pobeda)
 - Hostname             : srv2
 - IP Address           : <IP-Address2>
---------------------------------------------------------------
 - node ( master )      : <IP-Address1>
 - node ( replica )     : <IP-Address2>
 - arbiter              : <IP-Address3>
 - Manager version      : 2.1.13
 - Pangolin version     : 6.7.0
===============================================================
[student@srv2 ~]$ sudo -u postgres grep -B3 -n 'ssl: on' /etc/pangolin-manager/postgres.yml
226-        wal_sender_backup_directory: /pgdata/06/wal_backup
227-        wal_sender_check_crc: on
228-        wal_sender_check_type: block
229:        ssl: on

Останавливаем pangolin-manager и возвращаемся на мастер:

[student@srv2 ~]$ sudo systemctl stop pangolin-manager
[student@srv2 ~]$ exit
выход
Connection to srv2 closed.

Рестартуем pangolin-manager на мастере:

[student@srv1 ~]$ sudo systemctl restart pangolin-manager
[student@srv1 ~]$ sudo systemctl status pangolin-manager
Вывод команды:

● pangolin-manager.service - Runners to orchestrate a high-availability PostgreSQL
     Loaded: loaded (/usr/lib/systemd/system/pangolin-manager.service; enabled; preset: disabled)
    Drop-In: /usr/lib/systemd/system/service.d
             └─10-timeout-abort.conf
     Active: active (running) since Fri 2026-09-18 22:00:58 MSK; 6s ago
    Process: 50121 ExecStartPre=/bin/mkdir -p /var/run/postgresql (code=exited, status=0/SUCCESS)
    Process: 50122 ExecStartPre=/bin/chown -R postgres:postgres /var/run/postgresql (code=exited, status=0/SUCCESS)
   Main PID: 50123 (pangolin-manage)
      Tasks: 21 (limit: 4667)
     Memory: 200.3M
        CPU: 5.045s
     CGroup: /system.slice/pangolin-manager.service
             ├─50123 /opt/pangolin-manager/bin/pangolin-manager-bin/pangolin-manager.bin /etc/pangolin-manager/postgres.yml
             ├─50137 /usr/pangolin-6.7/bin/postgres -D /pgdata/06/data/ --config-file=/pgdata/06/data/postgresql.conf --listen_addresses=0.0.0.0 >
             ├─50140 "postgres: Pobeda: logger "
             ├─50141 "postgres: Pobeda: checkpointer "
             ├─50142 "postgres: Pobeda: wal backup writer "
             ├─50143 "postgres: Pobeda: background writer "
             ├─50145 "postgres: Pobeda: idle sessions terminator "
             ├─50146 "postgres: Pobeda: authproc "
             ├─50147 "postgres: Pobeda: password policy cache "
             ├─50148 "postgres: Pobeda: integrity check launcher "
             ├─50149 "postgres: Pobeda: license checker "
             ├─50154 "postgres: Pobeda: patroni postgres 127.0.0.1(38538) idle"
             ├─50177 "postgres: Pobeda: walwriter "
             ├─50178 "postgres: Pobeda: autovacuum launcher "
             ├─50179 "postgres: Pobeda: pg_cron launcher "
             └─50180 "postgres: Pobeda: logical replication launcher "

Запускаем на реплике:

[student@srv1 ~]$ ssh srv2 sudo systemctl start pangolin-manager
[student@srv1 ~]$ ssh srv2 sudo systemctl status pangolin-manager
Вывод команды:

● pangolin-manager.service - Runners to orchestrate a high-availability PostgreSQL
     Loaded: loaded (/usr/lib/systemd/system/pangolin-manager.service; enabled; preset: disabled)
    Drop-In: /usr/lib/systemd/system/service.d
             └─10-timeout-abort.conf
     Active: active (running) since Fri 2026-09-18 22:01:41 MSK; 5s ago
    Process: 81976 ExecStartPre=/bin/mkdir -p /var/run/postgresql (code=exited, status=0/SUCCESS)
    Process: 81977 ExecStartPre=/bin/chown -R postgres:postgres /var/run/postgresql (code=exited, status=0/SUCCESS)
   Main PID: 81978 (pangolin-manage)
      Tasks: 19 (limit: 4667)
     Memory: 184.1M
        CPU: 5.037s
     CGroup: /system.slice/pangolin-manager.service
             ├─81978 /opt/pangolin-manager/bin/pangolin-manager-bin/pangolin-manager.bin /etc/pangolin-manager/postgres.yml
             ├─81993 /usr/pangolin-6.7/bin/postgres -D /pgdata/06/data/ --config-file=/pgdata/06/data/postgresql.conf --listen_addresses=0.0.0.0 --port=5433 --cluster_name=Pobeda --wal_level=replica --hot_standby=on --max_connections=111 --max_wal_senders=10 --max_prepared_transactions=3 --max_locks_per_transaction=64 --track_commit_timestamp=False --max_replication_slots=10 --max_worker_processes=32 --wal_log_hints=True --authentication_max_workers=16
             ├─81996 "postgres: Pobeda: logger "
             ├─81997 "postgres: Pobeda: checkpointer "
             ├─81998 "postgres: Pobeda: wal backup writer "
             ├─81999 "postgres: Pobeda: background writer "
             ├─82000 "postgres: Pobeda: startup recovering 000000100000000000000015"
             ├─82001 "postgres: Pobeda: idle sessions terminator "
             ├─82003 "postgres: Pobeda: integrity check launcher "
             ├─82004 "postgres: Pobeda: license checker "
             ├─82005 "postgres: Pobeda: authproc "
             ├─82006 "postgres: Pobeda: password policy cache "
             ├─82007 "postgres: Pobeda: walreceiver streaming 0/15000218"
             └─82012 "postgres: Pobeda: patroni postgres 127.0.0.1(36124) idle"

Проверим настройку параметров:

[student@srv1 ~]$ sudo -iu postgres psql -c '\dconfig wal_sender_*'
           List of configuration parameters
           Parameter           |         Value
-------------------------------+-----------------------
 wal_sender_backup_directory   | /pgdata/06/wal_backup
 wal_sender_backup_strict      | off
 wal_sender_check_crc          | on
 wal_sender_check_type         | block
 wal_sender_panic_on_crc_error | off
 wal_sender_timeout            | 1min
(6 rows)

Проверим содержимое каталога резервных копий:

[student@srv1 ~]$ sudo ls -l /pgdata/06/wal_backup
итого 16400
-rw------- 1 postgres postgres 16777216 Sep 21 13:20 000000100000000000000015_backup
-rw------- 1 postgres postgres    16384 Sep 21 13:20 000000100000000000000015_crc

Видно, что копия сегмента WAL и файл контрольных сумм находятся в каталоге.

Включение режима декодирования​

Режим декодирования включается установкой следующих настроек:

  • wal_sender_check_crc: on
  • wal_sender_check_type: record

Отредактируем файл конфигурации на мастере и реплике соответствующим образом:

[student@srv1 ~]$ sudo -u postgres grep -B3 -n 'ssl: on' /etc/pangolin-manager/postgres.yml
226-        wal_sender_backup_directory: /pgdata/06/wal_backup
227-        wal_sender_check_crc: on
228-        wal_sender_check_type: record
229:        ssl: on

То же самое на реплике:

[student@srv1 ~]$ ssh srv2 sudo -u postgres grep -B3 -e 'ssl:[[:space:]]on' -n /etc/pangolin-manager/postgres.yml
226-        wal_sender_backup_directory: /pgdata/06/wal_backup
227-        wal_sender_check_crc: on
228-        wal_sender_check_type: record
229:        ssl: on

Остановим pangolin-manager на реплике:

[student@srv1 ~]$ ssh srv2 sudo systemctl stop pangolin-manager
[student@srv1 ~]$ ssh srv2 sudo systemctl status pangolin-manager
Вывод команды:

○ pangolin-manager.service - Runners to orchestrate a high-availability PostgreSQL
     Loaded: loaded (/usr/lib/systemd/system/pangolin-manager.service; enabled; preset: disabled)
    Drop-In: /usr/lib/systemd/system/service.d
             └─10-timeout-abort.conf
     Active: inactive (dead) since Sun 2026-02-15 17:53:55 MSK; 17s ago
   Duration: 19h 52min 13.136s
    Process: 81976 ExecStartPre=/bin/mkdir -p /var/run/postgresql (code=exited, status=0/SUCCESS)
    Process: 81977 ExecStartPre=/bin/chown -R postgres:postgres /var/run/postgresql (code=exited, status=0/SUCCESS)
    Process: 81978 ExecStart=/opt/pangolin-manager/bin/pangolin-manager-bin/pangolin-manager.bin /etc/pangolin-manager/postgres.yml (code=exited, status=0/SUCCESS)
   Main PID: 81978 (code=exited, status=0/SUCCESS)
        CPU: 19h 14min 31.218s

Перезапустим pangolin-manager на мастере:

[student@srv1 ~]$ sudo systemctl restart pangolin-manager
[student@srv1 ~]$ sudo systemctl status pangolin-manager
Вывод команды:

● pangolin-manager.service - Runners to orchestrate a high-availability PostgreSQL
     Loaded: loaded (/usr/lib/systemd/system/pangolin-manager.service; enabled; preset: disabled)
    Drop-In: /usr/lib/systemd/system/service.d
             └─10-timeout-abort.conf
     Active: active (running) since Sun 2026-02-15 17:54:58 MSK; 5s ago
    Process: 68256 ExecStartPre=/bin/mkdir -p /var/run/postgresql (code=exited, status=0/SUCCESS)
    Process: 68257 ExecStartPre=/bin/chown -R postgres:postgres /var/run/postgresql (code=exited, status=0/SUCCESS)
   Main PID: 68258 (pangolin-manage)
      Tasks: 21 (limit: 4667)
     Memory: 200.3M
        CPU: 4.457s
     CGroup: /system.slice/pangolin-manager.service
             ├─68258 /opt/pangolin-manager/bin/pangolin-manager-bin/pangolin-manager.bin /etc/pangolin-manager/postgres.yml
             ├─68272 /usr/pangolin-6.7/bin/postgres -D /pgdata/06/data/ --config-file=/pgdata/06/data/postgresql.conf --listen_addresses=0.0.0.0 >
             ├─68275 "postgres: Pobeda: logger "
             ├─68276 "postgres: Pobeda: checkpointer "
             ├─68277 "postgres: Pobeda: wal backup writer "
             ├─68278 "postgres: Pobeda: background writer "
             ├─68280 "postgres: Pobeda: idle sessions terminator "
             ├─68281 "postgres: Pobeda: authproc "
             ├─68282 "postgres: Pobeda: password policy cache "
             ├─68283 "postgres: Pobeda: integrity check launcher "
             ├─68284 "postgres: Pobeda: license checker "
             ├─68289 "postgres: Pobeda: patroni postgres 127.0.0.1(36780) idle"
             ├─68302 "postgres: Pobeda: walwriter "
             ├─68303 "postgres: Pobeda: autovacuum launcher "
             ├─68304 "postgres: Pobeda: pg_cron launcher "
             └─68305 "postgres: Pobeda: logical replication launcher "

Запустим pangolin-manager на реплике:

[student@srv1 ~]$ ssh srv2 sudo systemctl start pangolin-manager
[student@srv1 ~]$ ssh srv2 sudo systemctl status pangolin-manager
Вывод команды:

● pangolin-manager.service - Runners to orchestrate a high-availability PostgreSQL
     Loaded: loaded (/usr/lib/systemd/system/pangolin-manager.service; enabled; preset: disabled)
    Drop-In: /usr/lib/systemd/system/service.d
             └─10-timeout-abort.conf
     Active: active (running) since Sun 2026-02-15 17:55:51 MSK; 4s ago
    Process: 112372 ExecStartPre=/bin/mkdir -p /var/run/postgresql (code=exited, status=0/SUCCESS)
    Process: 112373 ExecStartPre=/bin/chown -R postgres:postgres /var/run/postgresql (code=exited, status=0/SUCCESS)
   Main PID: 112374 (pangolin-manage)
      Tasks: 19 (limit: 4667)
     Memory: 184.3M
        CPU: 3.530s
     CGroup: /system.slice/pangolin-manager.service
             ├─112374 /opt/pangolin-manager/bin/pangolin-manager-bin/pangolin-manager.bin /etc/pangolin-manager/postgres.yml
             ├─112389 /usr/pangolin-6.7/bin/postgres -D /pgdata/06/data/ --config-file=/pgdata/06/data/postgresql.conf --listen_addresses=0.0.0.0 --port=5433 --cluster_name=Pobeda --wal_level=replica --hot_standby=on --max_connections=111 --max_wal_senders=10 --max_prepared_transactions=3 --max_locks_per_transaction=64 --track_commit_timestamp=False --max_replication_slots=10 --max_worker_processes=32 --wal_log_hints=True --authentication_max_workers=16
             ├─112392 "postgres: Pobeda: logger "
             ├─112393 "postgres: Pobeda: checkpointer "
             ├─112394 "postgres: Pobeda: wal backup writer "
             ├─112395 "postgres: Pobeda: background writer "
             ├─112396 "postgres: Pobeda: startup recovering 000000110000000000000016"
             ├─112397 "postgres: Pobeda: idle sessions terminator "
             ├─112398 "postgres: Pobeda: authproc "
             ├─112399 "postgres: Pobeda: password policy cache "
             ├─112401 "postgres: Pobeda: integrity check launcher "
             ├─112402 "postgres: Pobeda: license checker "
             ├─112403 "postgres: Pobeda: walreceiver streaming 0/16000218"
             └─112408 "postgres: Pobeda: patroni postgres 127.0.0.1(40370) idle"

Подключившись к СУБД, проверим настройки:

[student@srv1 ~]$ sudo -iu postgres psql -c '\dconfig wal_sender_*'
           List of configuration parameters
           Parameter           |         Value
-------------------------------+-----------------------
 wal_sender_backup_directory   | /pgdata/06/wal_backup
 wal_sender_backup_strict      | off
 wal_sender_check_crc          | on
 wal_sender_check_type         | record
 wal_sender_panic_on_crc_error | off
 wal_sender_timeout            | 1min
(6 rows)

Сравним содержимое каталога резервных копий WAL в режимах проверки контрольных сумм и декодирования:

[student@srv1 ~]$ sudo ls -l /pgdata/06/wal_backup
итого 49172
-rw------- 1 postgres postgres 16777216 Sep 21 13:22 000000100000000000000015_backup
-rw------- 1 postgres postgres    16384 Sep 21 13:22 000000100000000000000015_crc
-rw------- 1 postgres postgres 16777216 Sep 21 13:22 000000100000000000000016_backup
-rw------- 1 postgres postgres        8 Sep 21 13:22 000000100000000000000016_crc
-rw------- 1 postgres postgres 16777216 Sep 21 13:22 000000110000000000000016_backup

Обратите внимание на то, как отличаются размеры файлов 000000100000000000000015_crc, созданного в режиме проверки контрольных сумм, и 000000100000000000000016_crc, созданного при работе в режиме декодирования.

Включение гибридного режима​

Гибридный режим включается при следующих настройках:

  • wal_sender_check_crc: on
  • wal_sender_check_type: 'record, block'

Отредактируем файлы конфигурации:

[student@srv1 ~]$ sudo -u postgres grep -B3 -e 'ssl:[[:space:]]on' -n /etc/pangolin-manager/postgres.yml
226-        wal_sender_backup_directory: /pgdata/06/wal_backup
227-        wal_sender_check_crc: on
228-        wal_sender_check_type: 'record, block'
229:        ssl: on

На реплике:

[student@srv1 ~]$ ssh srv2 sudo -u postgres grep -B3 -e 'ssl:[[:space:]]on' -n /etc/pangolin-manager/postgres.yml
226-        wal_sender_backup_directory: /pgdata/06/wal_backup
227-        wal_sender_check_crc: on
228-        wal_sender_check_type: 'record, block'
229:        ssl: on

Останавливаем реплику:

[student@srv1 ~]$ ssh srv2 sudo systemctl stop pangolin-manager

Рестартуем мастер:

[student@srv1 ~]$ sudo systemctl restart pangolin-manager
[student@srv1 ~]$ sudo systemctl status pangolin-manager
Вывод команды:

● pangolin-manager.service - Runners to orchestrate a high-availability PostgreSQL
     Loaded: loaded (/usr/lib/systemd/system/pangolin-manager.service; enabled; preset: disabled)
    Drop-In: /usr/lib/systemd/system/service.d
             └─10-timeout-abort.conf
     Active: active (running) since Sun 2026-02-15 18:48:43 MSK; 4s ago
    Process: 69212 ExecStartPre=/bin/mkdir -p /var/run/postgresql (code=exited, status=0/SUCCESS)
    Process: 69214 ExecStartPre=/bin/chown -R postgres:postgres /var/run/postgresql (code=exited, status=0/SUCCESS)
   Main PID: 69215 (pangolin-manage)
      Tasks: 21 (limit: 4667)
     Memory: 200.2M
        CPU: 3.577s
     CGroup: /system.slice/pangolin-manager.service
             ├─69215 /opt/pangolin-manager/bin/pangolin-manager-bin/pangolin-manager.bin /etc/pangolin-manager/postgres.yml
             ├─69229 /usr/pangolin-6.7/bin/postgres -D /pgdata/06/data/ --config-file=/pgdata/06/data/postgresql.conf --listen_addresses=0.0.0.0 >
             ├─69232 "postgres: Pobeda: logger "
             ├─69233 "postgres: Pobeda: checkpointer "
             ├─69234 "postgres: Pobeda: wal backup writer "
             ├─69235 "postgres: Pobeda: background writer "
             ├─69237 "postgres: Pobeda: idle sessions terminator "
             ├─69238 "postgres: Pobeda: authproc "
             ├─69239 "postgres: Pobeda: password policy cache "
             ├─69242 "postgres: Pobeda: integrity check launcher "
             ├─69243 "postgres: Pobeda: license checker "
             ├─69246 "postgres: Pobeda: patroni postgres 127.0.0.1(37818) idle"
             ├─69258 "postgres: Pobeda: walwriter "
             ├─69259 "postgres: Pobeda: autovacuum launcher "
             ├─69260 "postgres: Pobeda: pg_cron launcher "
             └─69261 "postgres: Pobeda: logical replication launcher "

Запускаем реплику:

[student@srv1 ~]$ ssh srv2 sudo systemctl start pangolin-manager
[student@srv1 ~]$ ssh srv2 sudo systemctl status pangolin-manager
Вывод команды:

● pangolin-manager.service - Runners to orchestrate a high-availability PostgreSQL
     Loaded: loaded (/usr/lib/systemd/system/pangolin-manager.service; enabled; preset: disabled)
    Drop-In: /usr/lib/systemd/system/service.d
             └─10-timeout-abort.conf
     Active: active (running) since Sun 2026-02-15 18:49:35 MSK; 4s ago
    Process: 113957 ExecStartPre=/bin/mkdir -p /var/run/postgresql (code=exited, status=0/SUCCESS)
    Process: 113958 ExecStartPre=/bin/chown -R postgres:postgres /var/run/postgresql (code=exited, status=0/SUCCESS)
   Main PID: 113959 (pangolin-manage)
      Tasks: 19 (limit: 4667)
     Memory: 184.2M
        CPU: 3.765s
     CGroup: /system.slice/pangolin-manager.service
             ├─113959 /opt/pangolin-manager/bin/pangolin-manager-bin/pangolin-manager.bin /etc/pangolin-manager/postgres.yml
             ├─113974 /usr/pangolin-6.7/bin/postgres -D /pgdata/06/data/ --config-file=/pgdata/06/data/postgresql.conf --listen_addresses=0.0.0.0 --port=5433 --cluster_name=Pobeda --wal_level=replica --hot_standby=on --max_connections=111 --max_wal_senders=10 --max_prepared_transactions=3 --max_locks_per_transaction=64 --track_commit_timestamp=False --max_replication_slots=10 --max_worker_processes=32 --wal_log_hints=True --authentication_max_workers=16
             ├─113977 "postgres: Pobeda: logger "
             ├─113978 "postgres: Pobeda: checkpointer "
             ├─113979 "postgres: Pobeda: wal backup writer "
             ├─113980 "postgres: Pobeda: background writer "
             ├─113981 "postgres: Pobeda: startup recovering 000000120000000000000017"
             ├─113982 "postgres: Pobeda: idle sessions terminator "
             ├─113983 "postgres: Pobeda: authproc "
             ├─113984 "postgres: Pobeda: password policy cache "
             ├─113986 "postgres: Pobeda: integrity check launcher "
             ├─113987 "postgres: Pobeda: license checker "
             ├─113988 "postgres: Pobeda: walreceiver streaming 0/17000218"
             └─113993 "postgres: Pobeda: patroni postgres 127.0.0.1(40994) idle"

Проверим настройки:

[student@srv1 ~]$ sudo -iu postgres psql -c '\dconfig wal_sender_*'
           List of configuration parameters
           Parameter           |         Value
-------------------------------+-----------------------
 wal_sender_backup_directory   | /pgdata/06/wal_backup
 wal_sender_backup_strict      | off
 wal_sender_check_crc          | on
 wal_sender_check_type         | record, block
 wal_sender_panic_on_crc_error | off
 wal_sender_timeout            | 1min
(6 rows)

Посмотрим, что изменилось в каталоге резервных копий WAL:

[student@srv1 ~]$ sudo ls -l /pgdata/06/wal_backup
итого 81956
-rw------- 1 postgres postgres 16777216 Sep 21 13:22 000000100000000000000015_backup
-rw------- 1 postgres postgres    16384 Sep 21 13:22 000000100000000000000015_crc
-rw------- 1 postgres postgres 16777216 Sep 21 13:22 000000100000000000000016_backup
-rw------- 1 postgres postgres        8 Sep 21 13:22 000000100000000000000016_crc
-rw------- 1 postgres postgres 16777216 Sep 21 13:22 000000110000000000000016_backup
-rw------- 1 postgres postgres 16777216 Sep 21 13:27 00000005000000000000000A_backup
-rw------- 1 postgres postgres 16777216 Sep 21 13:30 00000006000000000000000A_backup
-rw------- 1 postgres postgres    16384 Sep 21 13:30 00000006000000000000000A_crc

Заметно, что crc файл имеет такой же размер, как в режиме проверки контрольных сумм, так как в гибридном режиме они также проверяются.

Подведем итоги​

Вопрос 1

Вопрос 1: В каком файле конфигурации необходимо произвести изменения для присвоения значения параметру wal_sender_backup_directory?

Вопрос 2

Вопрос 2: Изменено значение параметра wal_sender_backup_directory в конфигурационном файле. Что надо сделать далее?