DR problems on some rendernodes

[tricky]

I have version 1.48.03 installed which, through this forum, I have been led to believe is pretty flawless - ie you can even stop renders and retsrt them without having to restart vrayspawner on all the nodes. However, I am still getting issues.

Granted, I don't seem to need to use the 'kill event' batch file anymore (at least I think that is the case), but I get random problems with different render nodes each time I set a DR going.

I have 9 Shuttle machines which are all identical spec, and 2 older dual XEON machines which are identical spec to eachother. I typically get at least 2 of the machines having the warning:
"warning: Render host shuttlexx (192.168.0.1xx) is not responding" where xx is the number of the machine.

Each time I submit, the problem machines change.

Any ideas what the problem is?

[Candelero]

Check Your network...

Best regards,
nikki Candelero

[tricky]

What kind of things should I look for/check? In a nutshell, we have a gigabit connection to each of the machines - surely this should be fast enough? Even if it wasn't, shouldn't vray just 'hold back' until sufficient bandwidth is available (assuming bandwidth is the issue)?

[tricky]

I also continue to get the mixed-up bucket problem where the only solution is to save, close and re-open max. This is a real pain esp. when working on a 50+MB max file. It seems DR is still quite some way from being resolved.

[jacksc02]

tricky, i am using the new DR in the .03 build and its running flawlessly.
Have you got the same build on every computer?

[tricky]

yes - it is the same on all machines. The email from Vlado told me to install this new version 'over the top' of the current install - is this what you did.

(I've got 2 high res renderes to get out and DR is just not working for me. The light cache goes OK (calculated on my local machine of course) but as soon as it then splits the render buckets to all my machines, they are all messed up! Bl**dy hell!)

[tricky]

heres a visual to explain the jumbled buckets:
correct, rendered traditionally using just one machine

jumbled, rendered using DR

[fabriceb]

Well, I have similar probelms with DR running 1.47.03. but just to say that I did loads and loads od DR with 1.46.?? until reccentlyand never had any problem whatsoever. Now I've upgraded and it's not as bad as this but buckets have different colour or tint, it seems to happen mostly in reflections.

[fabriceb]

Actually, looking at your image again, I remember I did get this problem often when I canceled the render early. The nodes get all mixed up. I reckon shut down all the machines and restart and try angain buy make sure you never cancel the render before all the nodes have started rendering properly, ie let them render 3 or 4 buckets and then cancel. Personnaly I always let the render finish when I do DR now....

[Da_elf]

i found that i got that when i had on motion blur and if there were nodes without the correct version. also i need to ask how many liscences you have? you listed 11 systems