Date: Fri, 28 Feb 2020 12:33:31 -0300
From: Alvaro Herrera <alvherre@2ndquadrant.com>
To: Peter Eisentraut <peter.eisentraut@2ndquadrant.com>
Cc: pgsql-hackers <pgsql-hackers@postgresql.org>
Subject: Re: Improve handling of parameter differences in physical replication
Message-ID: <20200228153331.GA31634@alvherre.pgsql>
MIME-Version: 1.0
Content-Type: text/plain; charset=iso-8859-1
Content-Disposition: inline
Content-Transfer-Encoding: 8bit
In-Reply-To: <4ad69a4c-cc9b-0dfe-0352-8b1b0cd36c7b@2ndquadrant.com>
User-Agent: Mutt/1.10.1 (2018-07-13)
Precedence: bulk

On 2020-Feb-27, Peter Eisentraut wrote:

> So this patch relaxes this a bit.  Upon receipt of
> XLOG_PARAMETER_CHANGE, we still check the settings but only issue a
> warning and set a global flag if there is a problem.  Then when we
> actually hit the resource issue and the flag was set, we issue another
> warning message with relevant information.  Additionally, at that
> point we pause recovery, so a hot standby remains usable.

Hmm, so what is the actual end-user behavior?  As I read the code, we
first send the WARNING, then pause recovery until the user resumes
replication; at that point we raise the original error.  Presumably, at
that point the startup process terminates and is relaunched, and replay
continues normally.  Is that it?

I think if the startup process terminates because of the original error,
after it is unpaused, postmaster will get that as a signal to do a
crash-recovery cycle, closing all existing connections.  Is that right?
If so, it would be worth improving that (possibly by adding a
sigsetjmp() block) to avoid the disruption.

Thanks,

-- 
Álvaro Herrera                https://www.2ndQuadrant.com/
PostgreSQL Development, 24x7 Support, Remote DBA, Training & Services