MIME-Version: 1.0
References: 
 <CAC0w7L+cxDkvgeM9vA_pPTAQ0pMo4PLuh9s-rCyR9AoLnYTL3A@mail.gmail.com>
In-Reply-To: 
 <CAC0w7L+cxDkvgeM9vA_pPTAQ0pMo4PLuh9s-rCyR9AoLnYTL3A@mail.gmail.com>
From: Jeff Janes <jeff.janes@gmail.com>
Date: Sat, 16 Apr 2022 00:37:10 -0400
Message-ID: 
 <CAMkU=1zL5+zQzzYL3j+xHBoGAw0JvmAoz_5c7+JZszybzAug5Q@mail.gmail.com>
Subject: Re: postgresql-14 slow query
To: Kenny Bachman <kenny.bachman17@gmail.com>
Cc: Pgsql-admin <pgsql-admin@lists.postgresql.org>
Content-Type: multipart/alternative; boundary="00000000000043e39305dcbe15f8"
Archived-At: 
 <https://www.postgresql.org/message-id/CAMkU%3D1zL5%2BzQzzYL3j%2BxHBoGAw0JvmAoz_5c7%2BJZszybzAug5Q%40mail.gmail.com>
Precedence: bulk

--00000000000043e39305dcbe15f8
Content-Type: text/plain; charset="UTF-8"

On Fri, Apr 15, 2022 at 4:59 PM Kenny Bachman <kenny.bachman17@gmail.com>
wrote:

> Hello Team,
>
> How can I tune this query? It got even slower when I created the index for
> (state_id, order_id desc). The following explain analyze output is
> without an index. It takes 13 seconds if I create that index. Could you
> help me?
>
> Thank you so much for your help.
>
> SELECT DISTINCT ON (order_history.order_id) order_id,
> order_history.creation_date  AS c_date
> FROM work.order_history WHERE (order_history.state_id = ANY (ARRAY[30, 51,
> 63, 136, 195, 233, 348])) AND order_history.is_false = 0
> ORDER BY order_history.order_id DESC;
>

This query doesn't make much sense to me.  You are selecting an arbitrary
creation_date for each order_id, which seems like a weird thing to do on
purpose.  Is your ORDER BY supposed to list another column in it to break
the ties?

I wonder how much benefit you are actually getting from the parallel
workers.  If you lower max_parallel_workers_per_gather, does the plan take
proportionally longer?

You should turn on track_io_timing, then repeat the query with EXPLAIN
(ANALYZE, BUFFERS).

Cheers,

Jeff

--00000000000043e39305dcbe15f8
Content-Type: text/html; charset="UTF-8"
Content-Transfer-Encoding: quoted-printable

<div dir=3D"ltr"><div dir=3D"ltr">On Fri, Apr 15, 2022 at 4:59 PM Kenny Bac=
hman &lt;<a href=3D"mailto:kenny.bachman17@gmail.com">kenny.bachman17@gmail=
.com</a>&gt; wrote:<br></div><div class=3D"gmail_quote"><blockquote class=
=3D"gmail_quote" style=3D"margin:0px 0px 0px 0.8ex;border-left:1px solid rg=
b(204,204,204);padding-left:1ex"><div dir=3D"ltr">Hello Team,<div><br></div=
><div>How can I tune this query? It got even slower when I created the inde=
x for (state_id, order_id desc). The following explain analyze output is wi=
thout=C2=A0an=C2=A0index. It takes 13 seconds if I create that index. Could=
 you help me?</div><div><br></div><div>Thank you so much for your help.</di=
v><div><br></div><div>SELECT DISTINCT ON (order_history.order_id) order_id,=
 order_history.creation_date =C2=A0AS c_date<br>FROM work.order_history WHE=
RE (order_history.state_id =3D ANY (ARRAY[30, 51, 63, 136, 195, 233, 348]))=
 AND order_history.is_false =3D 0<br>ORDER BY order_history.order_id DESC;<=
br></div></div></blockquote><div><br></div><div>This query doesn&#39;t make=
 much sense to me.=C2=A0 You are selecting an arbitrary creation_date for e=
ach order_id, which seems like a weird thing to do on purpose.=C2=A0 Is you=
r ORDER BY supposed to list another column in it to break the ties?</div><d=
iv><br></div><div>I wonder how much benefit you are actually getting from t=
he parallel workers.=C2=A0 If you lower max_parallel_workers_per_gather, do=
es the plan take proportionally longer?</div><div><br></div><div>You should=
 turn on track_io_timing, then repeat the query with EXPLAIN (ANALYZE, BUFF=
ERS).</div><div><br></div><div>Cheers,</div><div><br></div><div>Jeff</div><=
/div></div>

--00000000000043e39305dcbe15f8--