Horton

Re: [PATCH] Incremental sort (was: PoC: Partial sort)

James Coleman <jtc331@gmail.com>

View thread View on postgr.es
From: James Coleman <jtc331@gmail.com>
To: Justin Pryzby <pryzby@telsasoft.com>
Cc: Tomas Vondra <tomas.vondra@2ndquadrant.com>, Tom Lane <tgl@sss.pgh.pa.us>, Alvaro Herrera <alvherre@2ndquadrant.com>, Michael Paquier <michael@paquier.xyz>, Rafia Sabih <rafia.pghackers@gmail.com>, Peter Geoghegan <pg@bowt.ie>, Simon Riggs <simon@2ndquadrant.com>, Shaun Thomas <shaun.thomas@2ndquadrant.com>, Dmitry Dolgov <9erthalion6@gmail.com>, Alexander Korotkov <a.korotkov@postgrespro.ru>, PostgreSQL Developers <pgsql-hackers@lists.postgresql.org>, Peter Eisentraut <peter.eisentraut@2ndquadrant.com>, Andreas Karlsson <andreas@proxel.se>
Date: 2020-04-07T12:40:30Z
Lists: pgsql-hackers 
On Tue, Apr 7, 2020 at 12:25 AM Justin Pryzby <pryzby@telsasoft.com> wrote:
>
> On Mon, Apr 06, 2020 at 09:57:22PM +0200, Tomas Vondra wrote:
> > I've pushed the fist part of this patch series - I've reorganized it a
>
> I scanned through this again post-commit.  Find attached some suggestions.
>
> Shouldn't non-text explain output always show both disk *and* mem, including
> zeros ?

Could you give more context on this? Is there a standard to follow?
Regular sort nodes only ever report one type, so there's not a good
parallel there.

> Should "Pre-sorted Groups:" be on a separate line ?
> | Full-sort Groups: 1 Sort Method: quicksort Memory: avg=28kB peak=28kB Pre-sorted Groups: 1 Sort Method: quicksort Memory: avg=30kB peak=30kB

I'd originally had that, but Tomas wanted it to be more compact. It's
easy to adjust though if the consensus changes on that.

> And, should it use two spaces before "Sort Method", "Memory" and "Pre-sorted
> Groups"?  I think you should maybe do that instead of the "semicolon
> separator".  I think "two spaces" makes sense, since the units are different,
> similar to hash buckets and normal sort node.
>
>          "Buckets: %d  Batches: %d  Memory Usage: %ldkB\n",
>          appendStringInfo(es->str, "Sort Method: %s  %s: %ldkB\n",
>
> Note, I made a similar comment regarding two spaces for explain(WAL) here:
> https://www.postgresql.org/message-id/20200402054120.GC14618%40telsasoft.com
>
> And Peter E seemed to dislike that, here:
> https://www.postgresql.org/message-id/ef8c966f-e50a-c583-7b1e-85de6f4ca0d3%402ndquadrant.com

I read through that subthread, and the ending seemed to be Peter
wanting things to be unified. Was there a conclusion beyond that?

> Also, you're showing:
>         ExplainPropertyInteger("Maximum Sort Space Used", "kB",
>                                 groupInfo->maxMemorySpaceUsed, es);
>
> But in show_hash_info() and show_hashagg_info(), and in your own text output,
> that's called "Peak":
>         ExplainPropertyInteger("Peak Memory Usage", "kB", memPeakKb, es);
>         ExplainPropertyInteger("Peak Memory Usage", "kB",
>                                 spacePeakKb, es);

Yes, that's a miss and should be fixed.

James

Commits

  1. Further adjustments to Hashagg EXPLAIN ANALYZE output

  2. Rework EXPLAIN format for incremental sort

  3. Fix typos and improve incremental sort comments

  4. Stabilize incremental_sort tests

  5. Minor improvements in Incremental Sort explain

  6. Consider Incremental Sort paths at additional places

  7. Fix representation of SORT_TYPE_STILL_IN_PROGRESS.

  8. Fix failures in incremental_sort due to number of workers

  9. Fix show_incremental_sort_info with force_parallel_mode

  10. Implement Incremental Sort

  11. Fix handling of "Subplans Removed" field in EXPLAIN output.

  12. Fix EXPLAIN (SETTINGS) to follow policy about when to print empty fields.

  13. Ensure plpgsql result tuples have the right composite type marking.

  14. Propagate sort instrumentation from workers back to leader.

  15. Make new regression test case parallel-safe, and improve its output.

  16. Push limit through subqueries to underlying sort, where possible.

  17. Fix inappropriate printing of never-measured times in EXPLAIN.

  18. Fix some infelicities in EXPLAIN output for parallel query plans.