[pgpool-general: 5123] Re: PGPool failover backend communication error clarification

[Tatsuo Ishii]

Well, the user and the database settings for heal checking are fixed
and will not be changed frequently.  So an admin should notice the
configuration error at the earliest.  I don't think it's not worth the
trouble for taking care of careless admins.

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

> Unless I am mistaken, it seems to me that during health checks, if an
> ErrorResponse is returned by the server, a health check would fail.
> 
> I'm confused as to why this server behavior constitutes a backend
> communication error during health checks but not when a client wants to
> establish a connection to a server.
> 
> On Sun, Nov 13, 2016 at 6:39 PM Tatsuo Ishii <ishii at sraoss.co.jp> wrote:
> 
>> Hi,
>>
>> I have a question regarding the fail_over_on_backend_error option,
>> specifically what exactly constitutes a backend communication error.
>> Consider a client connecting to a server through pgpool. If for some
> reason
>> a SQL error is returned by the server before a SQL connection is
>> established (e.g. an ErrorResponse during start-up, see
>> https://www.postgresql.org/docs/current/static/protocol-flow.html) would
>> this constitute a backend communication error and mark the server as down?
> 
> No.
> 
>> I assume that this would be the case during normal health check
> procedures.
> 
> health check procedures are not affected by fail_over_on_backend_error.
> What's your concern with health check procedures?
> 
> Best regards,
> --
> Tatsuo Ishii
> SRA OSS, Inc. Japan
> English: http://www.sraoss.co.jp/index_en.php
> Japanese:http://www.sraoss.co.jp