Horton
Menu

Re: Add SPLIT PARTITION/MERGE PARTITIONS commands

Pavel Borisov <pashkin.elfe@gmail.com>

View thread
From: Pavel Borisov <pashkin.elfe@gmail.com>
To: Alexander Korotkov <aekorotkov@gmail.com>
Cc: Justin Pryzby <pryzby@telsasoft.com>, Dmitry Koval <d.koval@postgrespro.ru>, Alexander Lakhin <exclusion@gmail.com>, pgsql-hackers@lists.postgresql.org, Tomas Vondra <tomas.vondra@enterprisedb.com>, Alvaro Herrera <alvherre@alvh.no-ip.org>
Date: 2026-06-24T10:44:35Z
Lists: pgsql-hackers 
Hi, Alexander!

On Wed, 24 Jun 2026 at 14:25, Alexander Korotkov <aekorotkov@gmail.com> wrote:
>
> On Mon, Jun 22, 2026 at 1:44 PM Pavel Borisov <pashkin.elfe@gmail.com> wrote:
> > On Sun, 21 Jun 2026 at 16:11, Alexander Korotkov <aekorotkov@gmail.com> wrote:
> > >
> > > On Thu, Jun 18, 2026 at 6:49 PM Justin Pryzby <pryzby@telsasoft.com> wrote:
> > > > On Thu, Jun 18, 2026 at 10:31:11AM +0300, Alexander Korotkov wrote:
> > > > > Pushed with your suggestions accepted.
> > > >
> > > > Thanks.  When I went back to test this, I merged ~25 partitions that
> > > > were all on the same tablespace, but the merged table was created on the
> > > > default tablespace.
> > > >
> > > > I tried again with default_tablespace set, but it was ignored.  I think
> > > > that's wrong.  It's good to follow the tablespace of the parent table,
> > > > but if it has no tablespace set, default_tablespace should be obeyed.
> > > > See surrounding logic in DefineRelation.
> > > >
> > > > I see the docs say this:
> > > > +       <command>ALTER TABLE MERGE PARTITION</command> uses the partitioned
> > > > +       table itself as the template to construct the new partition.
> > > > +       The new partition will inherit the same table access method, persistence
> > > > +       type, and tablespace as the partitioned table.
> > >
> > > Correct, please see the attached patch, it makes
> > > createPartitionTable() deal with tablespaces the same was as
> > > DefineRelation() does.
> >
> > Thank you for working on this feature!
> > I've looked into the last v1 patch.
> >
> > Does it also worth inheriting DefineRelation()'s check and error for
> > the case if (tablespaceId == GLOBALTABLESPACE_OID)?
> > A brief look for default_tablespace GUC doesn't reveal a way why it
> > could not be set to global tablespace in a session.
>
> Thank you for catching this.  I've added this check and corresponding tests.

Looked at the patch v2.
I think in test we'd better split comments for the failing block (SET
default_tablespace TO pg_global;) and for the successful block (after
RESET default_tablespace;). Otherwise, the comment paragraph looks a
little hard to read to me:
+-- Parent has no explicit tablespace and default_tablespace is empty: the
+-- new partition uses the database default (reltablespace = 0).  Also
+-- exercise the pg_global rejection path with default_tablespace pointing
+-- at the shared tablespace.

Overall the patch looks good to me.

Regards,
Pavel Borisov
Supabase

Commits

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

  1. doc: clarify MERGE PARTITIONS adjacency requirement

  2. Take into account default_tablespace during MERGE/SPLIT PARTITION(S)

  3. Create TOAST table for partitions made by MERGE/SPLIT PARTITION

  4. Adjust errcode in checkPartition()

  5. Fix usage of palloc() in MERGE/SPLIT PARTITION(s) code

  6. Implement ALTER TABLE ... SPLIT PARTITION ... command

  7. Implement ALTER TABLE ... MERGE PARTITIONS ... command

  8. Calculate agglevelsup correctly when Aggref contains a CTE.

  9. Use PqMsg_* macros in applyparallelworker.c.

  10. Restrict psql meta-commands in plain-text dumps.

  11. Ensure we have a snapshot when updating various system catalogs.

  12. Use specific collation where needed in new test

  13. Expand virtual generated columns in the planner

  14. Virtual generated columns

  15. Define PG_LOGICAL_DIR for path pg_logical/ in data folder

  16. Revert support for ALTER TABLE ... MERGE/SPLIT PARTITION(S) commands

  17. Avoid repeated table name lookups in createPartitionTable()

  18. Provide deterministic order for catalog queries in partition_split.sql

  19. Don't copy extended statistics during MERGE/SPLIT partition operations

  20. Fix the name collision detection in MERGE/SPLIT partition operations

  21. Fix regression tests conflict in 3ca43dbbb6

  22. Add permission check for MERGE/SPLIT partition operations

  23. Fix one more portability shortcoming in new test_pg_dump test.

  24. Inherit parent's AM for partition MERGE/SPLIT operations

  25. Add tab completion for partition MERGE/SPLIT operations

  26. Rename tables in tests of partition MERGE/SPLIT operations

  27. Make new partitions with parent's persistence during MERGE/SPLIT

  28. Document the way partition MERGE/SPLIT operations create new partitions

  29. Change the way ATExecMergePartitions() handles the name collision

  30. Grammar fixes for split/merge partitions code

  31. Checks for ALTER TABLE ... SPLIT/MERGE PARTITIONS ... commands

  32. Fix some grammer errors from error messages and codes comments

  33. Support TZ and OF format codes in to_timestamp().

  34. Support identity columns in partitioned tables

  35. Fix indentation in twophase.c

  36. Fix corner-case planner failure for MERGE.

  37. Doc: fix documentation example for bytea hex output format.

  38. Avoid repeated name lookups during table and index DDL.