On Mon, 4 Aug 2003, Manfred Koizar wrote:
> On 01 Aug 2003 11:04:10 -0700, Jenny Zhang <jenny@osdl.org> wrote:
> >A sample OSDL-DBT3 test result report can be found at:
> >http://khack.osdl.org/stp/276912/
> >
> >Your comments are welcome,
>
> | effective_cache_size | 1000
>
> With 4GB of memory this is definitely too low and *can* (note that I
> don't say *must*) lead the planner to wrong decisions.
>
> | shared_buffers | 15200
>
> ... looks reasonable. Did you test with other values?
>
> | sort_mem | 524288
>
> This is a bit high, IMHO, but might be ok given that DBT3 is not run
> with many concurrent sessions (right?).
> http://khack.osdl.org/stp/276912/results/plot/vmstat_swap.png shows
> some swapping activity towards the end of the run which could be
> caused by a too high sort_mem setting.
Sorry, my last email shot off out of the gun before it was completed...
To repeat: Don't forget to set effective_cache_size the same way as
shared buffers (i.e. it's in 8k blocks for most systems.) If you have a
machine with 4 gig ram, and 3 gigs is available as disk cache, then
divide out 3 gigs by 8k to get the right number. My quick calculation
shows that being about 393216 blocks.
