Re: 035_standby_logical_decoding might fail due to FATAL message lost inside libpq

From: Alexander Lakhin <exclusion(at)gmail(dot)com>
To: "Hayato Kuroda (Fujitsu)" <kuroda(dot)hayato(at)fujitsu(dot)com>
Cc: pgsql-hackers <pgsql-hackers(at)postgresql(dot)org>
Subject: Re: 035_standby_logical_decoding might fail due to FATAL message lost inside libpq
Date: 2026-06-28 12:00:01
Message-ID: d373830a-afde-4c49-bb2a-35fbc650423b@gmail.com
Views: Whole Thread | Raw Message | Download mbox | Resend email
Thread:
Lists: pgsql-hackers

Hello Kuroda-san,

25.06.2026 05:04, Hayato Kuroda (Fujitsu) wrote:
> While sleeping, I came up with the idea; instead of waiting until the slot is
> active, we can wait pg_recvlogical receives the ack from the walsender and enters
> the loop. Attached patch implemented like that.
>
> On my env, the patch could resolve the issue. How do you feel?

Thank you for the fix!

I agree with your idea -- It makes sense to me to modify make_slot_active()
so that it returns the handle of pg_recvlogical after it reaches the
desired state from its own perspective.

With the the above repro modification applied, the fixed test passes for
me too.

Best regards,
Alexander

In response to

Browse pgsql-hackers by date

  From Date Subject
Next Message Jeevan Chalke 2026-06-28 12:56:02 Re: ON EMPTY clause for aggregate and window functions
Previous Message Henson Choi 2026-06-28 10:22:46 Re: Row pattern recognition