If jobs are updated on the primary server regularly and frequently, it will be easy for the query made by the reference job on the standby server to be canceled. In this case, edit one of the postgresql.conf file parameters shown in the table below.
Parameter | Description |
|---|---|
hot_standby_feedback | When "on" is set, the deletion (vacuum) of the data area that was deleted or updated on the primary server is suppressed. Accordingly, the query on the standby server will not be canceled. (*1) |
vacuum_defer_cleanup_age | The deletion (vacuum) of the data area that was deleted or updated on the primary server is delayed until the specified number of transactions is processed. Accordingly, the probability that the query on the standby server will be canceled decreases. |
*1: Because the vacuum is delayed, the data storage destination disk space of the primary server comes under pressure.
Additionally, if there is conflict between accesses and queries executed on the standby server, transaction logs indicating this conflict will be transferred.
Accordingly, specify as large a value as possible for the max_standby_archive_delay parameter so that access conflicts do not occur.
See
Refer to "Standby Servers" in the PostgreSQL Documentation for details on the hot_standby_feedback parameter.
Refer to "Master Server" in the PostgreSQL Documentation for details on the vacuum_defer_cleanup_age parameter.
