Horton

Re: [Proposal] Fully WAL logged CREATE DATABASE - No Checkpoints

Dilip Kumar <dilipbalaut@gmail.com>

View thread View on postgr.es
From: Dilip Kumar <dilipbalaut@gmail.com>
To: Justin Pryzby <pryzby@telsasoft.com>
Cc: Robert Haas <robertmhaas@gmail.com>, Tom Lane <tgl@sss.pgh.pa.us>, Andres Freund <andres@anarazel.de>, Ashutosh Sharma <ashu.coek88@gmail.com>, Maciek Sakrejda <m.sakrejda@gmail.com>, Bruce Momjian <bruce@momjian.us>, Alvaro Herrera <alvherre@alvh.no-ip.org>, Andrew Dunstan <andrew@dunslane.net>, Heikki Linnakangas <hlinnaka@iki.fi>, pgsql-hackers@lists.postgresql.org, Thomas Munro <thomas.munro@gmail.com>
Date: 2022-09-02T14:21:31Z
Lists: pgsql-hackers 
On Fri, Sep 2, 2022 at 5:25 PM Justin Pryzby <pryzby@telsasoft.com> wrote:
>
> On Tue, Aug 02, 2022 at 12:50:43PM -0500, Justin Pryzby wrote:
> > Also, if I understand correctly, this patch seems to assume that nobody is
> > connected to the source database.  But what's actually enforced is just that
> > nobody *else* is connected.  Is it any issue that the current DB can be used as
> > a source?  Anyway, both of the above problems are reproducible using a
> > different database.
> >
> > |postgres=# CREATE DATABASE new TEMPLATE postgres STRATEGY wal_log;
> > |CREATE DATABASE
>
> On Thu, Aug 04, 2022 at 05:16:04PM -0500, Justin Pryzby wrote:
> > On Thu, Aug 04, 2022 at 06:02:50PM -0400, Tom Lane wrote:
> > > The "invalidation" comment bothered me for awhile, but I think it's
> > > fine: we know that no other backend can connect to the source DB
> > > because we have it locked,
> >
> > About that - is it any problem that the currently-connected db can be used as a
> > template?  It's no issue for 2-phase commit, because "create database" cannot
> > run in an txn.
>
> Would anybody want to comment on this ?
> Is it okay that the *current* DB can be used as a template ?

I don't think there should be any problem with that.  The main problem
could have been that since we are reading the pg_class tuple block by
block there could be an issue if someone concurrently modifies the
pg_class or there are some tuples that are inserted by the prepared
transaction.  But in this case, the same backend can not have an open
prepared transaction while creating a database and that backend of
course can not perform any parallel operation as well.

-- 
Regards,
Dilip Kumar
EnterpriseDB: http://www.enterprisedb.com

Commits

  1. When using the WAL-logged CREATE DATABASE strategy, bulk extend.

  2. Avoid using a fake relcache entry to own an SmgrRelation.

  3. Fix data-corruption hazard in WAL-logged CREATE DATABASE.

  4. initdb: When running CREATE DATABASE, use STRATEGY = WAL_COPY.

  5. Simplify a needlessly-complicated regular expression.

  6. In 020_createdb.pl, change order of command-line arguments.

  7. Add new block-by-block strategy for CREATE DATABASE.

  8. Fix replay of create database records on standby

  9. Refactor code for reading and writing relation map files.

  10. Replace RelationOpenSmgr() with RelationGetSmgr().

  11. Refactor the fsync queue for wider use.