Horton

Re: Support logical replication of DDLs

Masahiko Sawada <sawada.mshk@gmail.com>

View thread
From: Masahiko Sawada <sawada.mshk@gmail.com>
To: Amit Kapila <amit.kapila16@gmail.com>
Cc: "Zhijie Hou (Fujitsu)" <houzj.fnst@fujitsu.com>, shveta malik <shveta.malik@gmail.com>, vignesh C <vignesh21@gmail.com>, Ajin Cherian <itsajin@gmail.com>, "Wei Wang (Fujitsu)" <wangw.fnst@fujitsu.com>, Runqi Tian <runqidev@gmail.com>, Peter Smith <smithpb2250@gmail.com>, Tom Lane <tgl@sss.pgh.pa.us>, li jie <ggysxcq@gmail.com>, Dilip Kumar <dilipbalaut@gmail.com>, Alvaro Herrera <alvherre@alvh.no-ip.org>, Japin Li <japinli@hotmail.com>, rajesh singarapu <rajesh.rs0541@gmail.com>, PostgreSQL Hackers <pgsql-hackers@lists.postgresql.org>, Zheng Li <zhengli10@gmail.com>
Date: 2023-04-26T06:31:13Z
Lists: pgsql-hackers

Commits

Same data as JSON: GET /api/v1/messages/:b64id/commits the thread's linked commits as JSON, with link sources. API reference →

  1. Add a run_as_owner option to subscriptions.

  2. Refactor pgoutput_change().

  3. Print the correct aliases for DML target tables in ruleutils.

  4. Fix object identity string for transforms

  5. Add grantable MAINTAIN privilege and pg_maintain role.

  6. Get rid of recursion-marker values in enum AlterTableType

  7. Release cache tuple when no longer needed

  8. Empty search_path in logical replication apply worker and walsender.

  9. Refactor format_type APIs to be more modular

  10. Use wrappers of PG_DETOAST_DATUM_PACKED() more.

On Wed, Apr 26, 2023 at 2:56 PM Amit Kapila <amit.kapila16@gmail.com> wrote:
>
> On Wed, Apr 26, 2023 at 10:01 AM Masahiko Sawada <sawada.mshk@gmail.com> wrote:
> >
> > On Tue, Apr 25, 2023 at 12:58 PM Zhijie Hou (Fujitsu)
> > <houzj.fnst@fujitsu.com> wrote:
> > >
> > > Aport from above comments, I splitted the code related to verbose
> > > mode to a separate patch. And here is the new version patch set.
> > >
> >
> > As for DDL replication, we create event triggers to write deparsed DDL
> > commands to WAL when creating a publication with the ddl option. The
> > event triggers are recreated/dropped at ALTER/DROP PUBLICATION. I'm
> > concerned it's possible that DDLs executed while such a publication
> > not existing are not replicated. For example, imagine the following
> > steps,
> >
> > 1. CREATE PUBLICATION test_pub ... WITH (ddl = 'table);
> > 2. CREATE SUBSCRIPTION test_sub ... PUBLICATION test_pub;
> > 3. ALTER SUBSCRIPTION test_sub DISABLE;
> > 4. DROP PUBLICATION test_pub;
> > 5. CREATE PUBLICATION test_pub ... WITH (ddl = 'table);
> > 6. ALTER SUBSCRIPTION test_sub ENABLE;
> >
> > DDLs executed between 4 and 5 won't be replicated.
> >
>
> But we won't even send any DMLs between 4 and 5. In fact, WALSender
> will give an error for those DMLs that publication doesn't exist as it
> uses a historic snapshot.

You're right, I missed this point.

> So, why do we expect DDLs between Drop and
> Create of publication should be replicated?

For example, suppose that a publication is created for a table and
then a new column is added to the table between 4 and 5, subsequent
INSERTs could fail due to the missing column. But it's not a problem
as you pointed out since the user dropped the publication.

Regards,

-- 
Masahiko Sawada
Amazon Web Services: https://aws.amazon.com