[pgpool-general: 7520] Re: Recovery after failure on a single backend

[Tatsuo Ishii]

> In our application we support both a clustered setup and a setup with
> a single node. In the situation with a single node, we still put
> pgpool between the application and the database for consistency. This
> means pgpool runs as a standalone node, with no watchdog and only a
> single backend. This works fine most of the time, however, if for some
> reason the database (or the connection to it) fails, pgpool marks the
> backend as invalid and never recovers. This will make a small hickup
> in the network to cause permanent outage of the service until pgpool
> is restarted. Is there any way to make pgpool reconnect to the
> database in this setup to recover from this situation?

There are two ways:

1) Enable disallow failover flag.

backend_flag0 = 'DISALLOW_TO_FAILOVER'

Upon failure to the backend, you will be getting following error:
psql: error: FATAL:  failed to create a backend connection
DETAIL:  executing failover on backend

Once the backend (or connection to the backend) comes up again, you
can connect to pgpool again. In this case failover never completes.

2) Enable auto_failback. Unlike #1, failover completes. But after some
period, the backend automatically becomes online again if connection
to the backend becomes healthy again.

Best regards,
--
Tatsuo Ishii
SRA OSS, Inc. Japan
English: http://www.sraoss.co.jp/index_en.php
Japanese:http://www.sraoss.co.jp