Playing with Elephants ~PostgreSQL and Pgpool-II~

Pgpool-II can monitor the status of PostgreSQL and automatically perform failover or degeneration if a failure is detected.  A healthy standby PostgreSQL server may be detached from Pgpool-II due to a temporary network problem. To re-attach a detached standby PostgreSQL server, you need to make sure that it is a healthy standby server and then manually run pcp_attach_node command or restart Pgpool-II.  Since Pgpool-II 4.1, it is possible to automatically re-attach a healthy standby PostgreSQL server by enabling auto_failback . In this post, I'll describe how to enable the automatic failback feature. Automatic failback To use this automatic failback feature, Pgpool-II's health check and streaming replication check must be enabled. Because Pgpool-II periodically retrieves information from pg_stat_replication to check the connectivity between primary and standby servers, PostgreSQL 9.1 or later is required. Pgpool-II will attach the PostgreSQL server, if the following condition