DR problem

[Rhys]

Ok so heres the setup

We have about 10 machines connected through a network, all with max and vray installed.

Ideally i want to be able to choose some free machines and dr to them. I find when i choose to do this i have the following problems.

When i set the render to irr map and qmc. say -3, -2 and have my machine and another machine with dr turned on. then dr works fine. the other prepass is done by the other machine and the image renders fine.

Now if i keep the same settings and say have 5 machines with dr on, what happens is that the irmap doesnt get saved properly and it renders frames all black or blue.

So i thought id calculate to irmap locally, then load it from file and have all the dr nodes pick up the irmap and render the buckets. but doing this did the same thing, the irmap didnt load on all nodes and the images render wrongly.

So makes me kinda give up on the dr front, im kind of stuck, its good for doing ppt renders as i can grab all the cores and render.

Is there anything im doing wrong, or does dr just not do what im trying to do with it.

Thanks !

[Rhys]

*bump*anyone ?

[palibebeh]

did you check that the other machines have access to the saved IRR map?
is the folder with the irr files shared across all the network?

[Rhys]

yes all the machines can access the same network drive z:

Is there any configuration to do for each render node?

[ene.xis]

Don' t use maped drives!
use UNC paths instead:
\\server\maps\

[palibebeh]

hmm

did You add all the folders/shared drives to user paths in max on all the nodes?

that's the thing that I forget about all the time, and then hello empty buckets, missing textures, proxies etc.

[Rhys]

but where would i link the unc paths to, as there is no files to be loaded.

As if i've saved my file and just render through the normal render window and not backburner. Where does the temp irrmap save itsself, so that the other machines can load it up.

It works fine when dr to one another machine, but not 2 machines or more.

[Rhys]

ok i made all paths that link to my xrefs to all unc paths.

and tired a little test, if i take my irrmap and lc that is saved and copy it to each machines locally to its C: the dr render works fine, all the machines pick up the save irmap and lc.

When i give the dr a unc path, and check the logs, it says that it can not load the irmap. and hence renders blank buckets.

one step closer, but not quite an ideal solution.

is there anything about configuring system paths to all be the same ?

[Rhys]

anyone?

[tehamnes]

hmm...It works here.

Check that the respective folders are shared across the network.

Oh, and did you perform one successful render on each slave through backburner first?

Hope you get it working