From: | Tom Lane <tgl(at)sss(dot)pgh(dot)pa(dot)us> |
---|---|
To: | Simon Riggs <simon(at)2ndquadrant(dot)com> |
Cc: | pgsql-patches(at)postgresql(dot)org |
Subject: | Re: effective_cache_size patch |
Date: | 2004-11-04 19:13:05 |
Message-ID: | 27927.1099595585@sss.pgh.pa.us |
Views: | Raw Message | Whole Thread | Download mbox | Resend email |
Thread: | |
Lists: | pgsql-patches |
Simon Riggs <simon(at)2ndquadrant(dot)com> writes:
> I enclose a doc patch for the effective_cache_size parameter in
> runtime.sgml: efcdoc.patch
Applied after translation into English ;-)
> Also, another minor patch which prevents effective_cache_size and
> random_page_cost from being set incorrectly: plancost.patch
> - previously it was possible to set effective_cache_size to 0, which
> would then be ignored and treated as 1 at run-time, so set minimum to 1
> and remove test at run-time.
I applied the guc.c change but did not remove the run-time test, just
to be on the safe side.
> - previously it was possible to set random_page_cost as a fractional
> value between 0 and 1, which screws up planner estimation, so set
> minimum value of 1 (i.e. random cost same as sequential cost).
Did not apply this. People sometimes use random_cost < 1 to force the
planner to pick index scans. Ideally this shouldn't be necessary...
regards, tom lane
From | Date | Subject | |
---|---|---|---|
Next Message | Tom Lane | 2004-11-04 19:21:45 | Re: [PATCHES] CVS should die (was: Possible make_oidjoins_check ...) |
Previous Message | Tom Lane | 2004-11-04 19:08:57 | Re: REINDEX doc patch |