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: Greg Nancarrow <gregn4422@gmail.com>
Cc: Robert Haas <robertmhaas@gmail.com>, Andres Freund <andres@anarazel.de>, Heikki Linnakangas <hlinnaka@iki.fi>, PostgreSQL Hackers <pgsql-hackers@lists.postgresql.org>
Date: 2021-12-01T12:34:07Z
Lists: pgsql-hackers 
On Wed, Dec 1, 2021 at 12:07 PM Greg Nancarrow <gregn4422@gmail.com> wrote:
>
> On Thu, Nov 25, 2021 at 10:17 PM Dilip Kumar <dilipbalaut@gmail.com> wrote:
> >
> > Thanks for the review and many valuable comments, I have fixed all of
> > them except this comment (/* If we got a cancel signal during the copy
> > of the data, quit */) because this looks fine to me.  0007, I have
> > dropped from the patchset for now.  I have also included fixes for
> > comments given by John.
> >
>
> I found the following issue with the patches applied:
>
> A server crash occurs after the following sequence of commands:
>
> create tablespace tbsp1 location '<directory>/tbsp1';
> create tablespace tbsp2 location '<directory>/tbsp2';
> create database test1 tablespace tbsp1;
> create database test2 template test1 tablespace tbsp2;
> alter database test2 set tablespace tbsp1;
> checkpoint;
>
> The following type of message is seen in the server log:
>
> 2021-12-01 16:48:26.623 AEDT [67423] PANIC:  could not fsync file
> "pg_tblspc/16385/PG_15_202111301/16387/3394": No such file or
> directory

Thanks a lot for testing this. From the error, it seems like some of
the old buffer w.r.t. the previous tablespace is not dropped after the
movedb.  Actually, we are calling DropDatabaseBuffers() after copying
to a new tablespace and dropping all the buffers of this database
w.r.t the old tablespace.  But seems something is missing, I will
reproduce this and try to fix it by tomorrow.  I will also fix the
other review comments raised by you in the previous mail.

-- 
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.