Horton

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

Rafia Sabih <rafia.sabih@enterprisedb.com>

View thread View on postgr.es
From: Rafia Sabih <rafia.sabih@enterprisedb.com>
To: Ashutosh Bapat <ashutosh.bapat@enterprisedb.com>
Cc: Robert Haas <robertmhaas@gmail.com>, Rajkumar Raghuwanshi <rajkumar.raghuwanshi@enterprisedb.com>, pgsql-hackers <pgsql-hackers@postgresql.org>
Date: 2017-03-22T12:01:45Z
Lists: pgsql-hackers 
On Wed, Mar 22, 2017 at 3:19 PM, Ashutosh Bapat
<ashutosh.bapat@enterprisedb.com> wrote:
>>>>
>> In an attempt to test the geqo side of this patch, I reduced
>> geqo_threshold to 6 and set enable_partitionwise_join to to true and
>> tried following query, which crashed,
>>
>> explain select * from prt, prt2, prt3, prt32, prt4, prt42 where prt.a
>> = prt2.b and prt3.a = prt32.b and prt4.a = prt42.b and prt2.a > 1000
>> order by prt.a desc;
>>
>> Stack-trace for the crash is as follows,
>>
> Nice catch. When reparameterize_path_by_child() may be running in a
> temporary memory context while running in GEQO mode. It may add a new
> PPI to base relation all in the temporary context. In the next GEQO
> cycle, the ppilist will be clobbered since the temporary context is
> reset for each geqo cycle. The fix is to allocate PPI in the same
> memory context as the RelOptInfo similar to mark_dummy_rel().
>
> I also found another problem. In geqo, we never call
> generate_partition_wise_join_paths() which set cheapest paths for each
> child-join. Because of this cheapest_*_paths are never set for those
> rels, thus segfaulting in functions like sort_inner_and_outer() which
> use those.
>
> Here's patch fixing both the issues. Please let me know if it fixes
> the issues you are seeing.

I tested the applied patch, it is fixing the reported issue.

-- 
Regards,
Rafia Sabih
EnterpriseDB: http://www.enterprisedb.com/

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".