Thread: running bonnie++

Hey folks,

During Greg Smith's lecture last week I could have sworn I saw on the
screen at some point a really long command line for bonnie++ - with
all the switches he uses.

But checking his slides I don't see this.

Am I mis-remembering?

Can someone recommend the best way to run it?  What combination of switches?

thanks,
-Alan

--
“Mother Nature doesn’t do bailouts.”
         - Glenn Prickett

You should be able to get a good idea of the options from "man bonnie++". I've always just used the defaults with
bonnie++

Also, you'll find Gregs older notes are here

http://www.westnet.com/~gsmith/content/postgresql/pg-disktesting.htm

--- On Wed, 27/5/09, Alan McKay <alan.mckay@gmail.com> wrote:

> From: Alan McKay <alan.mckay@gmail.com>
> Subject: [PERFORM] running bonnie++
> To: pgsql-performance@postgresql.org
> Date: Wednesday, 27 May, 2009, 2:24 PM
> Hey folks,
>
> During Greg Smith's lecture last week I could have sworn I
> saw on the
> screen at some point a really long command line for
> bonnie++ - with
> all the switches he uses.
>
> But checking his slides I don't see this.
>
> Am I mis-remembering?
>
> Can someone recommend the best way to run it?  What
> combination of switches?
>
> thanks,
> -Alan
>
> --
> “Mother Nature doesn’t do bailouts.”
>          - Glenn Prickett
>
> --
> Sent via pgsql-performance mailing list (pgsql-performance@postgresql.org)
> To make changes to your subscription:
> http://www.postgresql.org/mailpref/pgsql-performance
>

On Wed, 27 May 2009, Alan McKay wrote:

> During Greg Smith's lecture last week I could have sworn I saw on the
> screen at some point a really long command line for bonnie++ - with
> all the switches he uses.

You're probably thinking of the one I showed for sysbench, showing how to
use it to run a true random seeks/second with client load and size as
inputs.  I never run anything complicated with bonnie++ because the main
things I'd like to vary aren't there in the stable version anyway.  I've
found its main value is to give an easy to replicate test result, and
adding more switches moves away from that.  If you want to tweak the
values extensively, you probably should start climbing the learning curve
for iozone or fio instead.

That said, the latest unstable bonnie++ finally includes some concurrency
features that make tweaking it more useful, Josh Berkus's "Whack-a-mole"
tutorial showed some example while I just mentioned it in passing.  I'm
waiting until the 2.0 version comes out before I start switching my
examples over to it, again because of vendor repeatibility concerns.

--
* Greg Smith gsmith@gregsmith.com http://www.gregsmith.com Baltimore, MD