[Babase] Re: Pounding dbserve.biology.duke.edu

Karl O. Pinc babase@www.eco.princeton.edu
Tue, 12 Oct 2004 21:16:42 -0500


On 2004.10.12 17:02 Hunter Matthews wrote:
> At first glance it appears to be IO bound.

Not unusual for a database, but then you'd be suprised
how often databases are cpu bound.  Anyhow, increasing
the shared buffers should reduce the I/O dramatically.

> 
> An ssh'd vmstat isn't going to be available on the departmental
> server,
(seeing as vmstat takes no input it's hard to see how it's a security
risk, but ok no problem just had to say that)
> HOWEVER I think there's a cron job that collects that type of info.
> I'm
> certainly willign to copy that info into your home dir.

That'd be great.  (And easist for you I imagine.)  Hope it has a
timestamp and keeps some history (I'll find out when you do it.)

> 
> 
> 
> On Tue, 2004-10-12 at 16:12, Karl O. Pinc wrote:
> > Hi,
> >
> > I'm running a conversion program that's pounding
> > dbserve.biology.duke.edu.  I'm interested in knowing
> > whether it's CPU or I/O bound.
> >
> > Best would be if you could give me a command on login that runs
> vmstat
> > 10 on dbserve (using authorized_keys
> > command="", from="" syntax) but I'll take what I can get.
> >
> > I don't really _have_ to have this but I think it could
> > be handy.

Karl <kop@meme.com>
Free Software:  "You don't pay back, you pay forward."
                  -- Robert A. Heinlein