Robert Haas robertmh...@gmail.com:
Do you by any chance have a bazillion databases in this cluster? Can
you do these?
select sum(1) from pg_database;
1555
select pg_relation_size('pg_database');
221184
select sum(pg_column_size(d.*)) from pg_database;
That gave me:
ERROR:
max_fsm_pages = 280
max_fsm_relations = 16
What does the last couple of lines from a 'vacuum analyze verbose'
say? I have max_fsm_pages = 400 and max_fsm_relations = 1500.
You can also try to lower random_page_cost to a lower value like 1.2
but I doubt this will help in your
Claus Guttesen kome...@gmail.com:
Would love to get some advice on how to change my conf settings / setup to
get better I/O performance.
Server Specs:
2x Intel Xeon Quad Core (@2 Ghz - Clovertown,L5335)
4GB RAM
4x Seagate 73GB SAS HDD 10k RPM - in RAID ( stripped and mirrored )
On Fri, Sep 4, 2009 at 08:54, Scott Otisscott.o...@intand.com wrote:
Robert Haas robertmh...@gmail.com:
Do you by any chance have a bazillion databases in this cluster? Can
you do these?
select sum(1) from pg_database;
1555
Note that there are two features in 8.4 specifically designed to
Does the planner know how to use indices to optimize these queries?
For reference, I was having SEVERE performance problems with the
following comparison in an SQL statement where mask was an integer:
select ... from where .. and (permission mask = permission)
This resulted in the
So is there anything I can do in 8.3 to help this? I have tried setting '
track_activities', 'track_counts' and 'autovacuum' to 'off' (which has reduced
CPU and I/O a bit) - but the stats collector process is still using up a good
deal of CPU and I/O - is there any way to turn stats collecting
Scott Otis scott.o...@intand.com wrote:
So is there anything I can do in 8.3 to help this? I have tried
setting 'track_activities', 'track_counts' and 'autovacuum' to 'off'
(which has reduced CPU and I/O a bit)
You're going to regret that very soon, unless you are *very* sure you
have
Karl,
For reference, I was having SEVERE performance problems with the
following comparison in an SQL statement where mask was an integer:
select ... from where .. and (permission mask = permission)
AFAIK, the only way to use an index on these queries is through
expression