sure, I've test the patch [1] using script provided before, and it seems ok
[1]
https://www.postgresql.org/message-id/CAA4eK1LwQAEPJMTwVe3UYODeNMkK2QHf-WZF5aXp5ZcjDRcrUA%40mail.gmail.com
On Wed, May 25, 2022 at 4:09 AM Amit Kapila <amit(dot)kapila16(at)gmail(dot)com> wrote:
> On Wed, May 25, 2022 at 9:32 AM operations i <ioperations(dot)c(at)gmail(dot)com>
> wrote:
>
>> During a detailed test , I've found the order of replication slot
>> creation and publication creation is not the key point , but after
>> replication slot creation , there should not be any insertion to the
>> table,so I wonder why the change ( my case insert) will lead to
>> src/backend/catalog/pg_publication:1040 GetSysCacheOid1() cache miss.
>>
>
> It is due to the reason that both slot creation and 'insert' are before
> "CREATE PUBLICATION". Without inserts, it won't try to decode anything
> before the publication is created. Can you test it with the patch provided
> in the email [1] and if possible join that thread for discussion on this
> topic.
>
> [1] -
> https://www.postgresql.org/message-id/CAA4eK1LwQAEPJMTwVe3UYODeNMkK2QHf-WZF5aXp5ZcjDRcrUA%40mail.gmail.com
>
> --
> With Regards,
> Amit Kapila.
>