[Pgpool-general] 3.1 documentation

[Tatsuo Ishii]

> Tatsuo,
> 
> I'm confused by this warning:
> 
> "If a second standby took over primary when the first standby has
> already taken over too, you would get bogus data from the second
> standby. We recommend not to plan this kind of configuration. "
> 
> I thought that pgpoolII was designed to attempt failover on one node at
> a time, starting with the lowest numerical node.  Under what
> circumstances would pgPool attempt to failover more than one node at a time?

No, pgpool always one node at a time(except in the case when using
follow_masater_command). The particular portion of the document thinks
about following scenario:

1) we have one primary node(A) and two standby nodes (B and C)
2) A goes down and B takes over primary.
3) However C still thinks that C's primary node is A.
4) B goes down.
5) C takes over primary.

In this scenario, if any database updation happens between #2 and #4,
C is missing the updation and has bogus data.

To avoid this situation, you can use follow_master_command, which is
new in 3.1.  Currently there's few explanation how to use it in the
docs. I'm going to write some tutorials how to use the new directive.
--
Tatsuo Ishii
SRA OSS, Inc. Japan
English: http://www.sraoss.co.jp/index_en.php
Japanese: http://www.sraoss.co.jp