public inbox for [email protected]
help / color / mirror / Atom feedFrom: Tatsuo Ishii <[email protected]>
To: [email protected]
Cc: [email protected]
Cc: [email protected]
Cc: [email protected]
Cc: [email protected]
Subject: Re: Add memory/disk usage for WindowAgg nodes in EXPLAIN
Date: Sat, 14 Sep 2024 17:12:01 +0900 (JST)
Message-ID: <[email protected]> (raw)
In-Reply-To: <CAExHW5vjV1=oLsX8Hw8+M3Vf2Fuurq8HPUa4Ms8mC5H5DZPjcA@mail.gmail.com>
References: <CAExHW5vRPRLvsZYLmNGcDLkPDWDHXGSWYjox-to-OsCVFETd3w@mail.gmail.com>
<[email protected]>
<CAExHW5vjV1=oLsX8Hw8+M3Vf2Fuurq8HPUa4Ms8mC5H5DZPjcA@mail.gmail.com>
>> > or the case when the last usage fit in memory but an earlier
>> > usage spilled to disk.
>>
>> In my understanding once tuplestore changes the storage type to disk,
>> it never returns to the memory storage type in terms of
>> tuplestore_get_stats. i.e. once state->usedDisk is set to true, it
>> never goes back to false. So the test case is not necessary.
>> David, am I correct?
>
> I understand that. I am requesting a testcase to test that same logic.
Maybe something like this? In the example below there are 2
partitions. the first one has 1998 rows and the second one has 2
rows. Assuming that work_mem is 64kB, the first one does not fit the
memory and spills to disk. The second partition fits memory. However as
state->usedDisk remains true, explain shows "Storage: Disk".
test=# explain (analyze,costs off) select sum(n) over(partition by m) from (SELECT n < 3 as m, n from generate_series(1,2000) a(n));
n QUERY PLAN
--------------------------------------------------------------------------------
-------------
WindowAgg (actual time=1.958..473328.589 rows=2000 loops=1)
Storage: Disk Maximum Storage: 65kB
-> Sort (actual time=1.008..1.277 rows=2000 loops=1)
Sort Key: ((a.n < 3))
Sort Method: external merge Disk: 48kB
-> Function Scan on generate_series a (actual time=0.300..0.633 rows=2
000 loops=1)
Planning Time: 0.069 ms
Execution Time: 474515.476 ms
(8 rows)
Best reagards,
--
Tatsuo Ishii
SRA OSS K.K.
English: http://www.sraoss.co.jp/index_en/
Japanese:http://www.sraoss.co.jp
view thread (2+ messages)
reply
Reply instructions:
You may reply publicly to this message via plain-text email
using any one of the following methods:
* Reply to all the recipients using the --to and --cc options:
reply via email
To: [email protected]
Cc: [email protected], [email protected], [email protected], [email protected], [email protected]
Subject: Re: Add memory/disk usage for WindowAgg nodes in EXPLAIN
In-Reply-To: <[email protected]>
* Save the following mbox file, import it into your mail client,
and reply-to-all from there: mbox
This inbox is served by agora; see mirroring instructions
for how to clone and mirror all data and code used for this inbox