From: | Amit Kapila <amit(dot)kapila16(at)gmail(dot)com> |
---|---|
To: | Euler Taveira <euler(dot)taveira(at)2ndquadrant(dot)com> |
Cc: | PostgreSQL Hackers <pgsql-hackers(at)lists(dot)postgresql(dot)org> |
Subject: | Re: deferred primary key and logical replication |
Date: | 2020-10-27 10:46:43 |
Message-ID: | CAA4eK1LuoNACjV5vk-w4BKz2cV-HqmBnoZxQ1AixJ5WCraDNGg@mail.gmail.com |
Views: | Raw Message | Whole Thread | Download mbox | Resend email |
Thread: | |
Lists: | pgsql-hackers |
On Sun, Oct 25, 2020 at 9:39 PM Euler Taveira
<euler(dot)taveira(at)2ndquadrant(dot)com> wrote:
>
> On Mon, 5 Oct 2020 at 08:34, Amit Kapila <amit(dot)kapila16(at)gmail(dot)com> wrote:
>>
>> On Mon, May 11, 2020 at 2:41 AM Euler Taveira
>> <euler(dot)taveira(at)2ndquadrant(dot)com> wrote:
>> >
>> > Hi,
>> >
>> > While looking at an old wal2json issue, I stumbled on a scenario that a table
>> > with a deferred primary key is not updatable in logical replication. AFAICS it
>> > has been like that since the beginning of logical decoding and seems to be an
>> > oversight while designing logical decoding.
>> >
>>
>> I am not sure if it is an oversight because we document that the index
>> must be non-deferrable, see "USING INDEX records the old values of the
>> columns covered by the named index, which must be unique, not partial,
>> not deferrable, and include only columns marked NOT NULL." in docs
>> [1].
>>
>
> Inspecting this patch again, I forgot to consider that RelationGetIndexList()
> is called by other backend modules. Since logical decoding deals with finished
> transactions, it is ok to use a deferrable primary key.
>
But starting PG-14, we do support logical decoding of in-progress
transactions as well.
--
With Regards,
Amit Kapila.
From | Date | Subject | |
---|---|---|---|
Next Message | Bruce Momjian | 2020-10-27 10:53:08 | Re: Prevent printing "next step instructions" in initdb and pg_upgrade |
Previous Message | Peter Eisentraut | 2020-10-27 10:42:28 | Re: "unix_socket_directories" should be GUC_LIST_INPUT? |