Re: [PERFORM] regarding CLUSTER and HUGE work_mem / maintenance_work_mem

2012-01-27 Thread Scott Marlowe
On Fri, Jan 27, 2012 at 7:34 PM, Jon Nelson wrote: > On Fri, Jan 27, 2012 at 12:05 PM, Heikki Linnakangas > wrote: >> On 27.01.2012 19:43, Jon Nelson wrote: >>> >>> Let's say I have a 7GB table with 3-4 indices for a total of 10-12GB. >>> Furthermore, let's say I have a machine with sufficient me

Re: [PERFORM] regarding CLUSTER and HUGE work_mem / maintenance_work_mem

2012-01-27 Thread Jon Nelson
On Fri, Jan 27, 2012 at 12:05 PM, Heikki Linnakangas wrote: > On 27.01.2012 19:43, Jon Nelson wrote: >> >> Let's say I have a 7GB table with 3-4 indices for a total of 10-12GB. >> Furthermore, let's say I have a machine with sufficient memory for me >> to set the work_mem  and maintenance_work_mem

Re: [PERFORM] regarding CLUSTER and HUGE work_mem / maintenance_work_mem

2012-01-27 Thread Heikki Linnakangas
On 27.01.2012 19:43, Jon Nelson wrote: Let's say I have a 7GB table with 3-4 indices for a total of 10-12GB. Furthermore, let's say I have a machine with sufficient memory for me to set the work_mem and maintenance_work_mem to 20GB (just for this session). When I issue a CLUSTER using one of the

[PERFORM] regarding CLUSTER and HUGE work_mem / maintenance_work_mem

2012-01-27 Thread Jon Nelson
Let's say I have a 7GB table with 3-4 indices for a total of 10-12GB. Furthermore, let's say I have a machine with sufficient memory for me to set the work_mem and maintenance_work_mem to 20GB (just for this session). When I issue a CLUSTER using one of the indices, I see PostgreSQL (by way of str