Re: BUG #15888: Bogus "idle in transaction" state for logicaldecoding client after creating a slot - Mailing list pgsql-bugs

From Andres Freund
Subject Re: BUG #15888: Bogus "idle in transaction" state for logicaldecoding client after creating a slot
Date
Msg-id 20190710075501.6cpzzm35j6hj2ezv@alap3.anarazel.de
Whole thread Raw
In response to Re: BUG #15888: Bogus "idle in transaction" state for logicaldecoding client after creating a slot  (Michael Paquier <michael@paquier.xyz>)
List pgsql-bugs
Hi,

On 2019-07-10 13:51:47 +0900, Michael Paquier wrote:
> On Tue, Jul 09, 2019 at 05:46:30PM -0400, Alvaro Herrera wrote:
> > Hmm ... so what that commit did is precisely to fix this bug.
> > Magnus thought at the time he was fixing a pg10 bug,
> > https://postgr.es/m/CABUevEwX4g8y=gmgfPzxFKS7gqjSYNR949Xc96OQm=YXJmh_Og@mail.gmail.com
> > but apparently now we see that the bug was older than that.  Maybe it's
> > okay to backpatch further?
> 
> Yes, I would not be against a back-patch in this case.  There is a
> minor conflict because pre-9.6 WAL senders cannot run SQL queries but
> that looks simple enough to solve.

I'm not worried about backpatching that in isolation - but I'm worried
that just backpatching that indidividual commit isn't going to yield
particularly satisfactory behaviour. The code around this isn't super
robust, and pg10+ are noticably different due to the changes needed to
allow to execute normal queries over a replication connection.

Greetings,

Andres Freund



pgsql-bugs by date:

Previous
From: Michael Paquier
Date:
Subject: Re: BUG #15899: Valgrind detects errors on create gist index
Next
From: David Rowley
Date:
Subject: Re: PG11 - Multiple Key Range Partition