lschiere at users.sf.net
Wed Dec 20 08:09:59 EST 2006
On Wed, Dec 20, 2006 at 07:56:53AM -0500, Luke Schierer wrote:
> On Tue, Dec 19, 2006 at 05:47:19PM -0500, Ethan Blanton wrote:
> > Luke Schierer spake unto us the following wisdom:
> > > It takes me about half an hour to do the initial pull. I tested this in
> > > a new directory with a new mtn database to verify. time produced the
> > > following results:
> > To where are you pulling? Initial pull to Purdue (I2) was closer to
> > 50 minutes.
> > That said, pull is partially processor-bound; was our slice on the
> > server increased, or was the machine you pThe server slice has not
> > increased. That being said, its also not a serious limit, we are one
> > of 2-4 vhosts on a quad proc machine with a massive amount of ram.
> I was at work, so I have a relatively fast connection, and a relatively
> fast dual proc machine. I also have access to mrtg graphs of our network
> usage, and I was not bound by that at all. I'm timing the pull at the
> house, a verizon dsl connection, now. That is pulling betweek 40k and
> 95k. I'm fairly sure my connection can do more.
I stand corrected. Looking at the top in the server for a minute or so
of my pull, I see it sometimes spiking to 99% cpu usage. I was correct
in that memory is not the limit, though it is using 16%. That's a much
higher cpu percentage (more than double) what I see here, though it is
roughly the same amount of memory.
More information about the Cabal