Re: [PATCH] pg_hba.conf error messages for logical replication connections

From: "Euler Taveira" <euler(at)eulerto(dot)com>
To: "Amit Kapila" <amit(dot)kapila16(at)gmail(dot)com>, "Paul Martinez" <paulmtz(at)google(dot)com>
Cc: pgsql-hackers <pgsql-hackers(at)postgresql(dot)org>
Subject: Re: [PATCH] pg_hba.conf error messages for logical replication connections
Date: 2021-02-18 00:27:56
Message-ID: b18665b3-57fb-404d-ae39-4bebeabe540d@www.fastmail.com
Views: Raw Message | Whole Thread | Download mbox | Resend email
Thread:
Lists: pgsql-hackers

On Wed, Feb 17, 2021, at 8:01 AM, Amit Kapila wrote:
> Before fix
> ERROR: could not connect to the publisher: connection to server at
> "localhost" (::1), port 5432 failed: FATAL: pg_hba.conf rejects
> replication connection for host "::1", user "KapilaAm", no encryption
>
> After fix error:
> ERROR: could not connect to the publisher: connection to server at
> "localhost" (::1), port 5432 failed: FATAL: pg_hba.conf rejects
> connection for host "::1", user "KapilaAm", database "postgres", no
> encryption
> DETAIL: Logical replication connections do not match pg_hba.conf
> rules using the "replication" keyword.
The new message is certainly an improvement because it provides an additional
component (database name) that could be used to figure out what's wrong with
the logical replication connection. However, I wouldn't like to add a DETAIL
message for something that could be easily inspected in the pg_hba.conf. The
old message leaves a doubt about which rule was used (absence of database name)
but the new message makes this very clear. IMO with this new message, we don't
need a DETAIL message. If in doubt, user can always read that documentation
(the new sentence clarifies the "replication" usage for logical replication
connections).

Regarding the documentation, I think the new sentence a bit confusing. The
modified sentence is providing detailed information about "replication" in the
database field then you start mentioned "replication=database". Even though it
is related to the connection string, it could confuse the reader for a second.
I would say "it does not match logical replication connections". It seems
sufficient to inform the reader that he/she cannot use records with
"replication" to match logical replication connections.

--
Euler Taveira
EDB https://www.enterprisedb.com/

In response to

Responses

Browse pgsql-hackers by date

  From Date Subject
Next Message Greg Nancarrow 2021-02-18 01:03:11 Re: Parallel INSERT (INTO ... SELECT ...)
Previous Message Peter Smith 2021-02-18 00:17:54 Re: [HACKERS] logical decoding of two-phase transactions