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: 2022-03-14T16:34:04Z
Lists: pgsql-hackers 
On Mon, Mar 14, 2022 at 12:04 PM Robert Haas <robertmhaas@gmail.com> wrote:
> On Mon, Mar 14, 2022 at 7:51 AM Dilip Kumar <dilipbalaut@gmail.com> wrote:
> > Other than that, I have fixed some mistakes in comments and supported
> > tab completion for the new options.
>
> I was looking at 0001 and 0002 again and realized that I swapped the
> names load_relmap_file() and read_relmap_file() from what I should
> have done. Here's a revised version. With this, read_relmap_file() and
> write_relmap_file() become functions that just read and write the file
> without touching any global variables, and load_relmap_file() is the
> function that reads data from the file and puts it into a global
> variable, which seems more sensible than the way I had it before.

Regarding 0003 and 0005, I'm not a fan of 'bool isunlogged'. I think
'bool permanent' would be better (note BM_PERMANENT). This would
involve reversing true and false.

Regarding 0004, I can't really see a reason for this function to take
a LockRelId as a parameter rather than two separate OIDs. I also can't
entirely see why it should be called LockRelationId. Maybe
LockRelationInDatabaseById(Oid dbid, Oid relid, LOCKMODE lockmode)?
Note that neither caller actually has a LockRelId available; both have
to construct one.

Regarding 0005:

+       CREATEDB_WAL_LOG = 0,
+       CREATEDB_FILE_COPY = 1

I still think you don't need = 0 and = 1 here.

I'll probably go through and do a pass over the comments once you post
the next version of this. There seems to be work needed in a bunch of
places, but it probably makes more sense for me to go through and
adjust the things that seem to need it rather than listing a bunch of
changes for you to make.

-- 
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.