Date: Wed, 19 Oct 2022 12:02:20 +0200
From: hubert depesz lubaczewski <depesz@depesz.com>
Sender: depesz@depesz.com
To: Amit Kapila <amit.kapila16@gmail.com>
Cc: Kyotaro Horiguchi <horikyota.ntt@gmail.com>, pgsql-bugs@postgresql.org
Subject: Re: WAL segments removed from primary despite the fact that logical
 replication slot needs it.
Message-ID: <Y0/LLNYxyL30dfuV@depesz.com>
Reply-To: depesz@depesz.com
References: 
 <CAA4eK1JTE0tLgR5qzVECrKJV5VVTnDa0k5LQ4TOkBAA=MySj3g@mail.gmail.com>
 <20221017.155307.3782702886583901.horikyota.ntt@gmail.com>
 <CAA4eK1Jc6Q2p=L36iEFgKYW1V2Rz3nKRpJdtmhg60za5bBK5WQ@mail.gmail.com>
 <20221019.110032.2063205133677772338.horikyota.ntt@gmail.com>
 <CAA4eK1KXz=we0KfC441YJZRqbQnPUu5iSSV1Bh9em39Uat-tXA@mail.gmail.com>
MIME-Version: 1.0
Content-Type: text/plain; charset=utf-8
Content-Disposition: inline
In-Reply-To: 
 <CAA4eK1KXz=we0KfC441YJZRqbQnPUu5iSSV1Bh9em39Uat-tXA@mail.gmail.com>
Archived-At: 
 <https://www.postgresql.org/message-id/Y0/LLNYxyL30dfuV%40depesz.com>
Precedence: bulk

On Wed, Oct 19, 2022 at 03:16:59PM +0530, Amit Kapila wrote:
> Yeah, as requested yesterday, maybe checking if this can be reproduced
> will help us to narrow down the cause. If this doesn't happen due to
> an application mistake it appears that after the slot becomes inactive
> during the time walsender was down, the server removes the required
> files but there is nothing in the code that suggests that such a thing
> can happen.
> Are there any more logs on publisher/subscriber which indicate any
> sort of error?

This specific setup of replication was torn down, we couldn't keep it
running. Can setup it once more, and with some luck it will break again.
Will take a day, though.

Is there anything I can do to make it more debugable, but that will not
require restart of pg12?

Best regards,

depesz