Skip site navigation (1) Skip section navigation (2)

Re: Server Configuration

From: "Scott Marlowe" <scott(dot)marlowe(at)gmail(dot)com>
To: "Andy Dale" <andy(dot)dale(at)gmail(dot)com>
Cc: "pgsql general" <pgsql-general(at)postgresql(dot)org>
Subject: Re: Server Configuration
Date: 2008-05-20 17:42:50
Message-ID: dcc563d10805201042s7a610159xdcdb0de9ba6c60c8@mail.gmail.com (view raw or flat)
Thread:
Lists: pgsql-general
On Tue, May 20, 2008 at 8:14 AM, Andy Dale <andy(dot)dale(at)gmail(dot)com> wrote:
> Hi,
>
> I am currently trying to tweak Postgresql 8.1, to improve the overall
> performance of the database.  I have read over the following page/artical
> http://www.powerpostgresql.com/PerfList/, however at the moment (not for
> much longer hopefully) my hands aree tied when it comes to altering the
> kernel parameters, and thus allocating more shared buffers.
>
> I have read over the rest of the artical and adjusted some of the other
> values mentioned.  The values I have altered are:
>
> work_mem = 33554                            # min 64, size in KB

Depending on how many connections you're handling and how much memory
you have, this might be a little large, but it's not terrible.  Make
sure you aren't running your machine low on spare memory, as this can
cause the machine to start swapping and make it run slower.

> maintenance_work_mem = 33554        # min 1024, size in KB
> max_fsm_pages = 100000                  # min max_fsm_relations*16, 6 bytes
> each
>
> vacuum_cost_delay = 50                   # 0-1000 milliseconds

Maybe a little high.  most people find that 10 is just fine to keep
vacuum from slamming your I/O bandwidth.

> wal_buffers = 64                               # min 4, 8KB each
> commit_delay = 0                            # range 0-100000, in
> microseconds
> commit_siblings = 50                       # range 1-1000

Setting a short commit delay may allow for more siblings to get
committed together.

> effective_cache_size = 33333000     # typically 8KB each
>
> autovacuum = on                            # enable autovacuum subprocess?
> autovacuum_naptime = 30               # time between autovacuum runs, in
> secs
> autovacuum_vacuum_threshold = 400    # min # of tuple updates before vacuum
> autovacuum_analyze_threshold = 100    # min # of tuple updates before
> analyze
> autovacuum_vacuum_scale_factor = 0.4    # fraction of rel size before vacuum
> autovacuum_analyze_scale_factor = 0.2    # fraction of rel size before
> analyze
>
> Is it ok to have these settings with increasing the amount of shared buffers
> ?

Sure.  What's your max connections set to?

In response to

Responses

pgsql-general by date

Next:From: Decibel!Date: 2008-05-20 18:25:15
Subject: Short-circuiting FK check for a newly-added field
Previous:From: Karsten HilbertDate: 2008-05-20 17:35:44
Subject: Re: psql: FATAL: Ident authentication failedforuser"postgres"

Privacy Policy | About PostgreSQL
Copyright © 1996-2014 The PostgreSQL Global Development Group