Re: pgsql: Fix bug #16784 in Disk-based Hash Aggregation. - Mailing list pgsql-committers

From Jeff Davis
Subject Re: pgsql: Fix bug #16784 in Disk-based Hash Aggregation.
Date
Msg-id 802a29a0961c2ef0bf411cab535f5349741450e7.camel@j-davis.com
Whole thread Raw
In response to Re: pgsql: Fix bug #16784 in Disk-based Hash Aggregation.  (Jeff Davis <pgsql@j-davis.com>)
List pgsql-committers
On Sun, 2020-12-27 at 09:59 -0800, Jeff Davis wrote:
> I think I just need to disable sort. Because there is a () group, we
> don't need a Sort in the plan to exercise the bug.

That still didn't stabilize the test, so I removed the EXPLAIN part of
the test and left the execution test in place.

The planning for grouping sets still uses work_mem in the knapsack
algorithm, so something about the other platforms is not considering
the plan with all hashes (except the () grouping set). I think we can
tune the test to be stable, but it doesn't seem critical so I think
we're fine just doing the execution-time test.

Regards,
    Jeff Davis





pgsql-committers by date:

Previous
From: Jeff Davis
Date:
Subject: pgsql: Second attempt to stabilize 05c02589.
Next
From: Bruce Momjian
Date:
Subject: pgsql: Revert "Add key management system" (978f869b99) & later commits