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

Re: Less rows -> better performance?

From: Mario Weilguni <mweilguni(at)sime(dot)com>
To: Andreas Hartmann <andreas(at)apache(dot)org>
Cc: pgsql-performance(at)postgresql(dot)org
Subject: Re: Less rows -> better performance?
Date: 2008-07-21 15:29:14
Message-ID: (view raw, whole thread or download thread mbox)
Lists: pgsql-performance
Andreas Hartmann schrieb:
> Mario Weilguni schrieb:
>> Andreas Hartmann schrieb:
> […]
>>> I just verified that the autovacuum property is enabled.
> […]
>> Did you have:
>> stats_start_collector = on
>> stats_block_level = on
>> stats_row_level = on
>> Otherwise autovacuum won't run IMO.
> Thanks for the hint! The section looks like this:
> stats_start_collector = on
> #stats_command_string = off
> #stats_block_level = off
> stats_row_level = on
> #stats_reset_on_server_start = off
> I'll check the logs if the vacuum really runs - as soon as I find them :)
> -- Andreas
You might want to use these entries in your config:
redirect_stderr = on
log_directory = 'pg_log'
log_filename = 'postgresql-%Y-%m-%d_%H%M%S.log'
log_rotation_age = 1d

Fit those to your needs, then you will find log entries in $PGDATA/pg_log/

And BTW, I was wrong, you just need to have stats_row_level=On, 
stats_block_level doesn't matter. But in fact it's simple, if you don't 
have 24x7 requirements type VACUUM FULL ANALYZE; and check if your DB 
becomes smaller, I really doubt you can have that much indizes that 27MB 
dumps might use 2.3 GB on-disk.

You can check this too:
select relname, relpages, reltuples, relkind
  from pg_class
where relkind in ('r', 'i')
order by relpages desc limit 20;

Will give you the top-20 tables and their sizes, 1 page is typically 
8KB, so you can cross-check if relpages/reltuples is completly off, this 
is a good indicator for table/index bloat.


In response to


pgsql-performance by date

Next:From: Craig JamesDate: 2008-07-21 17:24:03
Subject: Re: Perl/DBI vs Native
Previous:From: Andreas HartmannDate: 2008-07-21 15:14:27
Subject: Re: Less rows -> better performance?

Privacy Policy | About PostgreSQL
Copyright © 1996-2017 The PostgreSQL Global Development Group