MIME-Version: 1.0
From: Jim Vanns <jvanns@ilm.com>
Date: Wed, 31 Jul 2024 15:16:45 +0100
Message-ID: <CAH7vdhP=7WHj_kgzkinVkpBHGTP521KhVg2g_mPu6imB_kdABA@mail.gmail.com>
Subject: Suggestions to overcome 'multixact "members" limit exceeded' in
 temporary tables
To: PostgreSQL-development <pgsql-hackers@postgresql.org>
Content-Type: multipart/alternative; boundary="0000000000002b8b35061e8bbf21"
Archived-At: <https://www.postgresql.org/message-id/CAH7vdhP%3D7WHj_kgzkinVkpBHGTP521KhVg2g_mPu6imB_kdABA%40mail.gmail.com>
Precedence: bulk

--0000000000002b8b35061e8bbf21
Content-Type: text/plain; charset="UTF-8"

I've reached the limit of my understanding and attempts at correcting my
code/use of temporary tables in the face of multixact members and have come
to ask for your help! Here's a brief description of my software;

Pool of N connection sessions, persistent for the duration of the program
lifetime.
Upon each session initialisation, a set of CREATE TEMPORARY TABLE ON COMMIT
DELETE ROWS statements are made for bulk ingest.
Each session is acquired by a thread for use when ingesting data and
therefore each temporary table remains until the session is terminated
The thread performs a COPY <temp table> FROM STDIN in binary format
Then an INSERT INTO <main table> SELECT FROM <temp table> WHERE...

This has been working great for a while and with excellent throughput.
However, upon scaling up I eventually hit this error;

ERROR:  multixact "members" limit exceeded
DETAIL:  This command would create a multixact with 2 members, but the
remaining space is only enough for 0 members.
HINT:  Execute a database-wide VACUUM in database with OID 16467 with
reduced vacuum_multixact_freeze_min_age and
vacuum_multixact_freeze_table_age settings.

And it took me quite a while to identify that it appears to be coming from
the temporary table (the other 'main' tables were being autovacuumed OK) -
which makes sense because they have a long lifetime, aren't auto vacuumed
and shared by transactions (in turn).

I first attempted to overcome this by introducing an initial step of always
creating the temporary table before the copy (and using on commit drop) but
this lead to a terrible performance degradation.
Next, I reverted the above and instead I introduced a VACUUM step every
1000000 (configurable) ingest operations
Finally, I introduced a TRUNCATE step in addition to the occasional VACUUM
since the TRUNCATE allowed the COPY option of FREEZE.

The new overhead appears minimal until after several hours and again I've
hit a performance degradation seemingly dominated by the TRUNCATE.

My questions are;

1) Is the VACUUM necessary if I use TRUNCATE + COPY FREEZE (on the
temporary table)?
2) Is there really any benefit to using FREEZE here or is it best to just
VACUUM the temporary tables occasionally?
3) Is there a better way of managing all this!? Perhaps re-CREATING the TT
every day or something?

I understand that I can create a Linux tmpfs partition for a tablespace for
the temporary tables and that may speed up the TRUNCATE but that seems like
a hack and I'd rather not do it at all if it's avoidable.

Thanks for your help,

Jim

PS. PG version in use is 15.4 if that matters here

-- 
Jim Vanns
Principal Production Engineer
Industrial Light & Magic, London

--0000000000002b8b35061e8bbf21
Content-Type: text/html; charset="UTF-8"
Content-Transfer-Encoding: quoted-printable

<div dir=3D"ltr">I&#39;ve reached the limit of my understanding and attempt=
s at correcting my code/use of temporary tables in the face of=C2=A0multixa=
ct members and have come to ask for your help! Here&#39;s a brief descripti=
on of my software;<div><br></div><div>Pool of N connection sessions, persis=
tent for the duration of the program lifetime.</div><div>Upon each session =
initialisation, a set of CREATE TEMPORARY TABLE ON COMMIT DELETE ROWS state=
ments are made for bulk ingest.</div><div>Each session is acquired by a thr=
ead for use when ingesting=C2=A0data and therefore each temporary table rem=
ains until the session is terminated</div><div>The thread performs a COPY &=
lt;temp table&gt; FROM STDIN in binary format</div><div>Then an INSERT INTO=
 &lt;main table&gt; SELECT FROM &lt;temp table&gt; WHERE...</div><div><br><=
/div><div>This has been working great for a while and with excellent throug=
hput. However, upon scaling up I eventually hit this error;</div><div><br><=
/div><div>ERROR: =C2=A0multixact &quot;members&quot; limit exceeded<br>DETA=
IL: =C2=A0This command would create a multixact with 2 members, but the rem=
aining space is only enough for 0 members.<br>HINT: =C2=A0Execute a databas=
e-wide VACUUM in database with OID 16467 with reduced vacuum_multixact_free=
ze_min_age and vacuum_multixact_freeze_table_age settings.<br></div><div><d=
iv><br></div><div>And it took me quite a while to identify that it appears =
to be coming from the temporary table (the other &#39;main&#39; tables were=
 being autovacuumed=C2=A0OK) - which makes sense because they have a long l=
ifetime, aren&#39;t auto vacuumed and shared by transactions (in turn).</di=
v><div><br></div><div>I first attempted to overcome this by introducing an =
initial step of always creating the temporary table before the copy (and us=
ing on commit drop) but this lead to a terrible performance degradation.</d=
iv><div>Next, I reverted the above and instead I introduced a VACUUM step e=
very 1000000 (configurable) ingest operations</div><div>Finally, I introduc=
ed a TRUNCATE step in addition to the occasional VACUUM since the TRUNCATE =
allowed the COPY option of FREEZE.</div><div><br></div><div>The new overhea=
d appears minimal until after several hours and again I&#39;ve hit a perfor=
mance degradation=C2=A0seemingly dominated=C2=A0by the TRUNCATE.</div><div>=
<br></div><div>My questions are;</div><div><br></div><div>1) Is the VACUUM =
necessary if I use TRUNCATE=C2=A0+ COPY FREEZE (on the temporary table)?</d=
iv><div>2) Is there really any benefit to using FREEZE here or is it best t=
o just VACUUM the temporary tables occasionally?</div><div>3) Is there a be=
tter way of managing all this!? Perhaps re-CREATING the TT every day or som=
ething?</div><div><br></div><div>I understand that I can create a Linux tmp=
fs partition for a tablespace for the temporary tables and that may speed u=
p the TRUNCATE but that seems like a hack and I&#39;d rather not do it at a=
ll if it&#39;s avoidable.</div><div><br></div><div>Thanks for your help,</d=
iv><div><br></div><div>Jim</div><div><br></div><div>PS. PG version in use i=
s 15.4 if that matters here</div><div><br></div><span class=3D"gmail_signat=
ure_prefix">-- </span><br><div dir=3D"ltr" class=3D"gmail_signature" data-s=
martmail=3D"gmail_signature"><div dir=3D"ltr">Jim Vanns<br>Principal Produc=
tion Engineer<br>Industrial Light &amp; Magic, London</div></div></div></di=
v>

--0000000000002b8b35061e8bbf21--