checkpoints are occurring too frequently

Started by Anibal David Acostaover 14 years ago4 messagesgeneral
Jump to latest
#1Anibal David Acosta
aa@devshock.com

I have a lot of entries like this in the log file

2011-11-17 02:02:46 PYST LOG: checkpoints are occurring too frequently (13
seconds apart)

2011-11-17 02:02:46 PYST HINT: Consider increasing the configuration
parameter "checkpoint_segments".

No, checkpoint parameters in postgres.conf are:

checkpoint_segments = 32 # in logfile segments, min
1, 16MB each

checkpoint_timeout = 10min # range 30s-1h

checkpoint_completion_target = 0.9 # checkpoint target duration, 0.0 -
1.0

What should be a correct value for checkpoint_segments to avoid excessive
checkpoint events?

#2Grzegorz Jaśkiewicz
gryzman@gmail.com
In reply to: Anibal David Acosta (#1)
Re: checkpoints are occurring too frequently

increase your checkpoint segments

--
GJ

#3Gabriele Bartolini
gabriele.bartolini@2ndQuadrant.it
In reply to: Anibal David Acosta (#1)
Re: checkpoints are occurring too frequently

Hi Anibal,

On Thu, 17 Nov 2011 09:48:10 -0300, "Anibal David Acosta"
<aa@devshock.com> wrote:

What should be a correct value for checkpoint_segments to avoid
excessive checkpoint events?

There is no golden rule or value that fits all scenarios. Usually 32 is
a good value to start with, however it might not be perfectly tailored
for your environment.

To give you an idea, currently you are issuing a checkpoint every
32*16MB of WAL traffic (or every 10 minutes). Maybe you can describe us
better your workload, if it is subject to usage spikes or regularly
distributed throughout the day mainly in terms of
inserts/updates/deletes.

I would gradually try and increase checkpoint_segments by 32 and
monitor the effects.

Cheers,
Gabriele

--
Gabriele Bartolini - 2ndQuadrant Italia
PostgreSQL Training, Services and Support
Gabriele.Bartolini@2ndQuadrant.it - www.2ndQuadrant.it

#4Anibal David Acosta
aa@devshock.com
In reply to: Gabriele Bartolini (#3)
Re: checkpoints are occurring too frequently

Thanks!

-----Mensaje original-----
De: Gabriele Bartolini [mailto:Gabriele.Bartolini@2ndQuadrant.it]
Enviado el: jueves, 17 de noviembre de 2011 10:14 a.m.
Para: Anibal David Acosta
CC: pgsql-general@postgresql.org
Asunto: Re: [GENERAL] checkpoints are occurring too frequently

Hi Anibal,

On Thu, 17 Nov 2011 09:48:10 -0300, "Anibal David Acosta"
<aa@devshock.com> wrote:

What should be a correct value for checkpoint_segments to avoid
excessive checkpoint events?

There is no golden rule or value that fits all scenarios. Usually 32 is a good value to start with, however it might not be perfectly tailored for your environment.

To give you an idea, currently you are issuing a checkpoint every 32*16MB of WAL traffic (or every 10 minutes). Maybe you can describe us better your workload, if it is subject to usage spikes or regularly distributed throughout the day mainly in terms of inserts/updates/deletes.

I would gradually try and increase checkpoint_segments by 32 and monitor the effects.

Cheers,
Gabriele

--
Gabriele Bartolini - 2ndQuadrant Italia
PostgreSQL Training, Services and Support
Gabriele.Bartolini@2ndQuadrant.it - www.2ndQuadrant.it