Re: pgbench --startup option - Mailing list pgsql-hackers

From Peter Geoghegan
Subject Re: pgbench --startup option
Date
Msg-id CAEYLb_Uv7kCgMqRCK4FZJJUcZt2Q0YxOEYQjD+LUfhaEAbW35A@mail.gmail.com
Whole thread Raw
In response to pgbench --startup option  (Jeff Janes <jeff.janes@gmail.com>)
List pgsql-hackers
On 10 February 2013 23:27, Jeff Janes <jeff.janes@gmail.com> wrote:
> While looking at some proposed patches and pondering some questions on
> performance, I realized I desperately needed ways to run benchmarks with
> different settings without needing to edit postgresql.conf and
> restart/reload the server each time.

I'd thought about hacking this capability into pgbench-tools, so that
there was a new outer-most loop that iterates through different
postgresql.conf files. Come to think of it, the need to vary settings
per test set (that is, per series of benchmarks, each of which is
plotted as a different color) generally only exists for one or two
settings, so it is probably better to pursue the approach that you
propose here.

I guess what I've outlined could still be useful for PGC_POSTMASTER
gucs, like shared_buffers.

-- 
Regards,
Peter Geoghegan



pgsql-hackers by date:

Previous
From: Jeff Janes
Date:
Subject: pgbench --startup option
Next
From: Tom Lane
Date:
Subject: Re: BUG #7493: Postmaster messages unreadable in a Windows console