public inbox for [email protected]  
help / color / mirror / Atom feed
From: Koshino Taiki <[email protected]>
To: Macao Tom <[email protected]>
Cc: [email protected] <[email protected]>
Subject: Re: About pgpool failover / switchover
Date: Wed, 17 Sep 2025 01:13:34 +0000
Message-ID: <TYCP286MB2926C37D1EB8D8E2D882DA819417A@TYCP286MB2926.JPNP286.PROD.OUTLOOK.COM> (raw)
In-Reply-To: <CAPTV4vzgri_2VogHSZUv9-h2+PkWKdAcESg3eOwxkf2MV7UrYQ@mail.gmail.com>
References: <CAPTV4vzgri_2VogHSZUv9-h2+PkWKdAcESg3eOwxkf2MV7UrYQ@mail.gmail.com>

This is because, when a failover or switchover occurs and the primary changes, Pgpool-II updates $PGDATA/myrecovery.conf instead of postgresql.auto.conf.

When a standby connects to the primary, it refers to the information in postgresql.auto.conf , if the setting exists there.
However, during failover or switchover, Pgpool-II updates $PGDATA/myrecovery.conf rather than postgresql.auto.conf.
As a result, postgresql.auto.conf still contains outdated connection information, causing PostgreSQL to attempt to connect using this wrong information, which leads to connection failures.

In this case, by deleting postgresql.auto.conf, the primary_conninfo in $PGDATA/myrecovery.conf was used, which was correct, so the connection succeeded.

Since having primary_conninfo in postgresql.auto.conf can sometimes prevent proper connections, it is generally recommended not to include primary_conninfo in postgresql.auto.conf.

Taiki Koshino<[email protected]>
SRA OSS K.K.
TEL: 03-5979-2701 FAX: 03-5979-2702
URL: https://www.sraoss.co.jp/

________________________________
差出人: Macao Tom <[email protected]>
送信日時: 2025年9月12日 12:43
宛先: [email protected] <[email protected]>
件名: About pgpool failover / switchover

Dear pgpool,

I built up a postgresql 17.6.2 3 nodes cluster with 3 pgpool 4.6.2 , when I try to switch over using “pcp_promote_start -s …..” it seems other standy nodes can not follow the new primary node. However, after I remove the “postgres.auto.conf” in each server and switch over using the same command again, it works. I have also check what store in “postgres.auto.conf”, it may store many config like “primary_conn_info” which connect to the old primary.

Do you have any idea about this? Or do you know what mechanism cause postgres store the old primary info in “postgres.auto.config” file?

Best regards,
Terence


reply

Reply instructions:

You may reply publicly to this message via plain-text email
using any one of the following methods:

* Reply to all the recipients using the --to and --cc options:
  reply via email

  To: [email protected]
  Cc: [email protected], [email protected], [email protected]
  Subject: Re: About pgpool failover / switchover
  In-Reply-To: <TYCP286MB2926C37D1EB8D8E2D882DA819417A@TYCP286MB2926.JPNP286.PROD.OUTLOOK.COM>

* Save the following mbox file, import it into your mail client,
  and reply-to-all from there: mbox

This inbox is served by agora; see mirroring instructions
for how to clone and mirror all data and code used for this inbox