Horton

Re: Partition-wise join for join between (declaratively) partitioned tables

Amit Langote <langote_amit_f8@lab.ntt.co.jp>

View thread View on postgr.es
From: Amit Langote <Langote_Amit_f8@lab.ntt.co.jp>
To: Ashutosh Bapat <ashutosh.bapat@enterprisedb.com>, Robert Haas <robertmhaas@gmail.com>
Cc: Rafia Sabih <rafia.sabih@enterprisedb.com>, Rajkumar Raghuwanshi <rajkumar.raghuwanshi@enterprisedb.com>, pgsql-hackers <pgsql-hackers@postgresql.org>
Date: 2017-03-24T10:48:39Z
Lists: pgsql-hackers

Attachments

Hi Ashutosh,

On 2017/03/23 21:48, Ashutosh Bapat wrote:
>>> I have fixed all the issues reported till now.

I've tried to fix your 0012 patch (Multi-level partitioned table
expansion) considering your message earlier on this thread [1].
Especially the fact that no AppendRelInfo and RelOptInfo are allocated for
partitioned child tables as of commit d3cc37f1d [2].  I've fixed
expand_inherited_rtentry() such that AppendRelInfo *is* allocated for a
partitioned child RTEs whose rte->inh is set to true.  Such an RTE is
recursively expanded with that RTE the parent.

Also as I mentioned elsewhere [3], the multi-level inheritance expansion
of partitioned table will break update/delete for partitioned table, which
is because inheritance_planner() is not ready to handle inheritance sets
structured that way.  I tried to refactor inheritance_planner() such that
its core logic can be recursively invoked for partitioned child RTEs.  The
resulting child paths and other auxiliary information related to planning
across the hierarchy are maintained in one place using a struct to hold
the same in a few flat lists.  The refactoring didn't break any existing
tests and a couple of new tests are added to check that it indeed works
for multi-level partitioned tables expanded using new multi-level structure.

There is some test failure in 0014 (Multi-level partition-wise join
tests), probably because of the changes I made to 0012, which I didn't get
time to check why, although I've checked using an example that multi-level
join planning still works, so it's not completely broken either.

Thanks,
Amit

[1]
https://www.postgresql.org/message-id/CAFjFpRefs5ZMnxQ2vP9v5zOtWtNPuiMYc01sb1SWjCOB1CT%3DuQ%40mail.gmail.com

[2] https://git.postgresql.org/gitweb/?p=postgresql.git;a=commit;h=d3cc37f1d

[3]
https://www.postgresql.org/message-id/744d20fe-fc7b-f89e-8d06-6496ec537b86%40lab.ntt.co.jp

Commits

  1. Basic partition-wise join functionality.

  2. Assorted preparatory refactoring for partition-wise join.

  3. Teach adjust_appendrel_attrs(_multilevel) to do multiple translations.

  4. Stamp 10beta2.

  5. Eat XIDs more efficiently in recovery TAP test.

  6. Abstract logic to allow for multiple kinds of child rels.

  7. Implement SortSupport for macaddr data type

  8. Attempt to stabilize grouping sets regression test plans.

  9. Teach xlogreader to follow timeline switches

  10. Don't scan partitioned tables.

  11. Fix grammar.

  12. postgres_fdw: Push down FULL JOINs with restriction clauses.

  13. Some preliminary refactoring towards partitionwise join.

  14. contrib/amcheck needs RecentGlobalXmin to be PGDLLIMPORT'ified.

  15. Print test parameters like "foo: 123", and results like "foo = 123".