On 10/17/2011 04:48 AM, Micka wrote:
Hi,I've a postgres 9.1 database used for map generating ( tiles ). The system has 24Go RAM and 5 processors. I'm using geoserver to generate the tiles. My data used 8486 MB => psql -d gis -c "SELECT pg_size_pretty(pg_database_size('gis'))" I've carefully indexes the table by the "the_geom" column. Here is my database config : --> change : --> listen_addresses = '*' --> max_connections = 50 --> tcp_keepalives_idle = 60 # TCP_KEEPIDLE, in seconds; --> shared_buffers = 1024MB # 10% of available RAM --> work_mem = 256MB # min 64kB --> maintenance_work_mem = 256MB # min 1MB --> effective_cache_size = 5120MB --> autovacuum = off sudo nano /etc/sysctl.conf --> kernel.shmmax=5368709120 --> kernel.shmall=5368709120 I wanted to have your opinion about this config ? What can I do to optimize the performance ? Thank you,
Yeah... We're gonna need some more details. Whats slow? Are you CPU bound or IO bound? How many concurrent db connections? What does vmstat look like? And 10% of 24 gig is 2.4 gig, not 1 gig. Or is this box doing something else. I noticeeffective_cache_size is only 5 gig, so you must be doing other things on this box.
--> autovacuum = off
Are you vacuuming by hand!? If not this is a "really bad idea" (tm)(c)(r) -Andy -- Sent via pgsql-performance mailing list (pgsql-performance@postgresql.org) To make changes to your subscription: http://www.postgresql.org/mailpref/pgsql-performance
