[Bf-funboard] Re: Daemon Integration

[Paul Lunneberg]

My only concern with network rendering is I would love to have it functional
across the internet and not just my local network.  But as you said the
default could be setup such that it would grab files off of a central
machine on the local network. Maybe just make it possible to set an FTP site
and path to the source files as well as an output directory.  That would
take care of my requirements... then each job could potentially ftp the
output files to different locations fairly easily.  You would still have the
scaling issue that you could potentially have a lot of systems downloading
the .blend files and textures every time... but obviously there would have
to be some sort of cache on the local system while a job is running as well
as a check each frame to see if it has the appropriate files.

As for the daemon being seperate... That works fine... I believe integrating
them as far as interface is concerned would make things a lot easier for
people to setup.  Especially if the interface to configure them is kept
simple and organized!

Paul Lunneberg

----- Original Message ----- 
From: "Roland Hess" <rolandh@reed-witting.com>
To: <bf-funboard@blender.org>
Sent: Wednesday, June 18, 2003 6:20 AM
Subject: [Bf-funboard] Re: Daemon Integration


> Nice mockup. My initial thought is that it would be better in most
> instances (i.e., you're only using this if you have a lot of
> rendering to do) to have the source files for your artwork (.blends
> and image maps) reside on a server. It could even be the main machine
> that you're working from. But the point is, I think the client
> machines should pull the .blend files and textures off of the server
> machine, as opposed to the render manager pushing the data. Perhaps
> another setting in the Client Settings section would be the network
> path to the source data. Also, completed frames would be directed in
> a similar fashion, instead of being returned to the render manager.
>
> Having the render manager perform these functions does not scale
> well. If there are only two or three client machines, it's fine.
> Let's say I get 20 PII's for $79 a piece and stick them in a closet
> in my basement. I don't want each one having to wait for the render
> manager to pump a 20 MB .blend file and 320 MB of textures (because
> there's avi's used as textures) to each machine in sequence, and then
> sit there and try to field the requests from all twenty machines as
> they send back frames. Much more efficient to set a network path,
> send the commands, then simply monitor the results.
> -- 
> Roland Hess
> harkyman