Horton

Re: Default setting for enable_hashagg_disk

Tomas Vondra <tomas.vondra@2ndquadrant.com>

View thread View on postgr.es
From: Tomas Vondra <tomas.vondra@2ndquadrant.com>
To: Tom Lane <tgl@sss.pgh.pa.us>
Cc: "David G. Johnston" <david.g.johnston@gmail.com>, Stephen Frost <sfrost@snowman.net>, Peter Geoghegan <pg@bowt.ie>, Jeff Davis <pgsql@j-davis.com>, Alvaro Herrera <alvherre@2ndquadrant.com>, Andres Freund <andres@anarazel.de>, Bruce Momjian <bruce@momjian.us>, Robert Haas <robertmhaas@gmail.com>, David Rowley <dgrowleyml@gmail.com>, Justin Pryzby <pryzby@telsasoft.com>, Melanie Plageman <melanieplageman@gmail.com>, "pgsql-hackers@postgresql.org" <pgsql-hackers@postgresql.org>
Date: 2020-07-12T12:30:43Z
Lists: pgsql-hackers, pgsql-docs 
On Sat, Jul 11, 2020 at 08:47:54PM -0400, Tom Lane wrote:
>Tomas Vondra <tomas.vondra@2ndquadrant.com> writes:
>> I don't know, but one of the main arguments against simply suggesting
>> people to bump up work_mem (if they're hit by the hashagg spill in v13)
>> was that it'd increase overall memory usage for them. It seems strange
>> to then propose a new GUC set to a default that would result in higher
>> memory usage *for everyone*.
>
>It seems like a lot of the disagreement here is focused on Peter's
>proposal to make hash_mem_multiplier default to 2.0.  But it doesn't
>seem to me that that's a critical element of the proposal.  Why not just
>make it default to 1.0, thus keeping the default behavior identical
>to what it is now?
>
>If we find that's a poor default, we can always change it later;
>but it seems to me that the evidence for a higher default is
>a bit thin at this point.
>

You're right, I was specifically pushing against that aspect of the
proposal. Sorry for not making that clearer, I assumed it's clear from
the context of this (sub)thread.

I agree making it 1.0 (or equal to work_mem, if it's not a multiplier)
by default, but allowing it to be increased if needed would address most
of the spilling issues.


regards

-- 
Tomas Vondra                  http://www.2ndQuadrant.com
PostgreSQL Development, 24x7 Support, Remote DBA, Training & Services

Commits

  1. Add hash_mem_multiplier GUC.

  2. HashAgg: use better cardinality estimate for recursive spilling.

  3. Remove hashagg_avoid_disk_plan GUC.

  4. Doc fixup for hashagg_avoid_disk_plan GUC.

  5. Rework HashAgg GUCs.

  6. Disk-based Hash Aggregation.

  7. Implement partition-wise grouping/aggregation.

  8. Defer creation of partially-grouped relation until it's needed.