From: Tom Lane <tgl@sss.pgh.pa.us>
To: Mladen Gogala <gogala.mladen@gmail.com>
cc: Pgsql Performance <pgsql-performance@lists.postgresql.org>
Subject: Re: XA transactions much slower on 14.2 than on 13.5
In-reply-to: <b6c413ee-19a4-f3a8-9772-210049c22369@gmail.com>
References: <b6c413ee-19a4-f3a8-9772-210049c22369@gmail.com>
Comments: In-reply-to Mladen Gogala <gogala.mladen@gmail.com>
	message dated "Fri, 04 Mar 2022 21:33:01 -0500"
MIME-Version: 1.0
Content-Type: text/plain; charset="UTF-8"
Content-ID: <4007943.1646448265.1@sss.pgh.pa.us>
Content-Transfer-Encoding: 8bit
Date: Fri, 04 Mar 2022 21:44:25 -0500
Message-ID: <4007944.1646448265@sss.pgh.pa.us>
Archived-At: 
 <https://www.postgresql.org/message-id/4007944.1646448265%40sss.pgh.pa.us>
Precedence: bulk

Mladen Gogala <gogala.mladen@gmail.com> writes:
> My boss asked me to upgrade one of the development  databases from 13.5 
> --> 14.2. One thing that we've noticed right away is that XA 
> transactions (2-phase commit) are much slower on 14.2 than on 13.5. Were 
> there any significant changes to the XA protocol in the version 14? Did 
> anybody else encountered this problem?

There were a bunch of changes around the 2PC code to support logical
replication of 2PC transactions, but I don't think they should have
made for any particular performance difference in non-replicated
servers.  Can you put together a self-contained test case that
demonstrates what you're seeing?

			regards, tom lane