Horton

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

Robert Haas <robertmhaas@gmail.com>

View thread View on postgr.es
From: Robert Haas <robertmhaas@gmail.com>
To: Dilip Kumar <dilipbalaut@gmail.com>
Cc: 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>, PostgreSQL Hackers <pgsql-hackers@lists.postgresql.org>
Date: 2022-03-23T13:33:34Z
Lists: pgsql-hackers 
On Wed, Mar 23, 2022 at 9:19 AM Dilip Kumar <dilipbalaut@gmail.com> wrote:
> I think directly using smgrcreate() is a better idea instead of first
> registering and then unregistering it.   I have made that change in
> the attached patch.  After this change now we can merge creating the
> MAIN_FORKNUM also in the loop below where we are creating other
> fork[1] with one extra condition but I think current code is in more
> sync with the other code where we are doing the similar things so I
> have not merged it in the loop.  Please let me know if you think
> otherwise.

Generally I think our practice is that we do the main fork
unconditionally (because it should always be there) and the others
only if they exist. I suggest that you make this consistent with that,
but you could do it like if (forkNum != MAIN_FORKNUM &&
!smgrexists(...)) continue if that seems nicer.

Do you think that this version handles pending syncs correctly? I
think perhaps that is overlooked.

-- 
Robert Haas
EDB: 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.