DR problems - actively refused etc

[tricky]

We are having some issues with getting DR working since building some additional render slaves, and I hope somebody could assist.

Our main server is running SBS2003 with a domain.
We are using Max2009 x64 and Vray2 on all machines.
3 workstations are running Windows 7 Ultimate.
8 original slaves are running XP64.
3 new slaves are also running XP64.

DR has been working flawlessly for a few years, and it is only since building these new machines that we have had any issues.

All the slaves have spawner running as a service with administrative rights. We tried to configure the 3 new machines in the same way, but when we try to start the service (with admin rights), we are given an error and the service will not start. The only way we can get the service to start on these machines was to have them use their local accounts (they are each logged in as an administrator). However, the problem is then that these machines have to be logged in in order to run the service. The only way we ficed this problem (the machines have no monitors, keyboards or mice connected to them) is to install TweekUI which automatically logs them in as an administrator. DR then works on these 3 new machines.

We now have a problem where the original 8 slaves do not render. When we send a DR job to them, we have:-

Warning: Could not connect to host 192.168.0.xxx: No connection could be made because the target machine actively refused it.

We can't be sure that having the new 3 machines has caused the problems, but we have literally only just finished building them so it seems something must be related. We have tried leaving the 3 new machines OFF and just using the original 8, but they all still give the above error message, even after rebooting them!

As with most of these sorts of problems, it is no doubt a combination of things that is preventing things from working smoothly.

Can anyone suggest something else to try? I am really starting to pull my hair out here!!!....

[tricky]

To simplify the problem, I have now solved the spawner/license issues on the 3 new slaves. It was a DNS error. Having now solved that, the remaining problem is the

Warning: Could not connect to host 192.168.0.xxx: No connection could be made because the target machine actively refused it.

error on my original 8 render slaves. I will keep digging around, but I am not sure what has changed on these.

[tricky]

Re-installed vray on the offending nodes and the problem seemed to go away. Good grief.

[plastic_]

I think the problem is xp64.
I'm running windows 7 on every box and I have no issues with DR.
It may also help that I'm mirroring most assets to local drives of the DR nodes, instead of using a fileserver for everything. This reduces load and makes things more stable.

[tricky]

I think the problem is xp64.
I'm running windows 7 on every box and I have no issues with DR.
It may also help that I'm mirroring most assets to local drives of the DR nodes, instead of using a fileserver for everything. This reduces load and makes things more stable.

I think we've solved it. It was a number of unrelated issues that were compounded and made the problem solving more difficult. We now have a fully functional render farm again!

Interesting point about the mirroring though. Are you using some kind of mirroring software on each render node to keep an exact copy of all the assets (textures/proxies/ies etc) on a local drive on each node? How do you then ensure each node is referencing the local assets rather than the server?

[plastic_]

I'm using syncback.
I have a folder, c:\textures\
actually it isn't on c:, but it's a "ntfs symbolic link folder". This allows me to put the files on any drive, but the path stays always the same.
I'm using this absolute path for my textures in max.
Whenever I add some texture, I start the syncback mirroring action and all render nodes will get syncronized.

[tricky]

I'm using syncback.
I have a folder, c:\textures\
actually it isn't on c:, but it's a "ntfs symbolic link folder". This allows me to put the files on any drive, but the path stays always the same.
I'm using this absolute path for my textures in max.
Whenever I add some texture, I start the syncback mirroring action and all render nodes will get syncronized.

I am familiar with syncback, though we are using Allwaysync at the moment. Do you find calculable time savings with this approach?

[flino2004]

I also have a copy of textures, proxies, etc locally in every node...it's a lot faster that way and I only have stock assets there, so if i have custom textures for a project I use a Map folder inside the project folder...If one of the custom textures will be used often in other projects I mirror it.
The new version of RefaMo has a feature to copy files in one shot to all the nodes.