Re: [dvd::rip] Cluster Mode System Requirements

On Wed, Aug 03, 2005 at 12:44:40PM -0400, Phillip Geiger wrote:

> How fast a computer do you really need to make a useful node in a
> dvd::rip cluster?
> 
> I'm planning to use a PIII 866 fileserver I've got as the primary
> workstation, host for the cluster control daemon, and NFS server. 
> Most of the time I'll have one or two 1.7 GHz P4 laptop plugged in to
> serve as a node.  But, I've also got about ten PII 333 MHz laptops
> that are idle much of the time.
> 
> The cluster mode docs page says:
>     "if the performance of your cluster nodes differs much you
>      can decrease [chunk size] to prevent slow nodes from
>      blocking the whole cluster with transcoding a huge chunk
>      while the others are idle. But: decreasing the chunk size
>      too much makes 2-pass encoding useless, because the
>      material for analysis becomes too short."
> 
> So, I'd probably want to whack the chunk size way down so the 1.7 GHz
> P4 doesn't sit idle while the slower machines chug along.  But, I
> don't want to mess up the 2-pass encoding.  My question is this -
> before I go to the trouble of setting up a whole bunch of underpowered
> nodes, at what point is a machine just _too slow_ to be a useful node?
>  Am I wasting my time with a bunch of 333 MHz dinosaurs?

I have an almost identical setup, so I'm really interested in the answer to
this question as well.

Also, I noticed in the docs that it urges against using the NFS "soft" mount
option, but on my wireless LAN, NFS almost always hangs without the soft
option.  I can't even kill -9 any process that tries to access the NFS drive
once that happens, and I have to reboot.  Is there any way around this?