Thnks for the replyies. It's a slony slave db, for reporting. So, what's a good value to set to effective_cache_size with 10 Gb RAM?
On Thu, Jul 9, 2009 at 9:05 PM, Scott Mead <[email protected]>wrote: > > On Thu, Jul 9, 2009 at 6:18 PM, Rafael Domiciano < > [email protected]> wrote: > >> Hello People, >> >> Today, I've upgraded a dedicated postgres server, from 2 Gb to 10 Gb. >> Everything gone well. >> >> But, I would like shared buffers to use at least 5 Gb of the total memory. >> > > What's your workload? Is this db primarily for reporting or OLTP? > > If you have an OLTP style workload, I wouldn't recommend going much > over 2.5 - 4 GB (depending on your specific workload). Just set your > 'effective_cache_size' higher. This tells postgres how much memory that the > OS has for caching and the database will perform better. > >> Linux Fedora Core 9 >> postgres=# select version(); >> version >> >> --------------------------------------------------------------------------------------------------- >> PostgreSQL 8.3.5 on i686-pc-linux-gnu, compiled by GCC gcc (GCC) 4.3.0 >> 20080428 (Red Hat 4.3.0-8) >> (1 row) >> > > 32 bit pg can't address that much memory. You'd need to recompile or > download the 64 bit packages. I believe you'd need to dump / reload as > well, but I may be off about that one. > > > --Scott > >
