I have a small problem with C4D Team Render and TurbulenceFD:
even though I set the simulation cache to a path on a mapped drive and every render node can see the path, the workstation seems to store the complete cache again locally during team rendering and distributes it directly to the nodes… which unfortunately is quite slow… (the log of the render nodes says “Downloading Assets” although the render cache should be read directly from the mapped drive).
Am I missing anything?
I am not sure about this as I do not use DR but correct me if I am wrong …the cache would have to be “downloaded” to each machine from the central store as each machine will need to load the cache into it’s GPU or CPU memory. If the cache is not local it by definition would have to be downloaded locally. I think “Downloading Assets” is each node downloading the cache from the mapped drive and of course an scene assets on the master workstation, GI caches shadow caches etc
Thanks for your input! I digged a bit deeper and checked the user folders while submitting the job to team-rendering… Cinema copied ALL the cache from the mapped drive to the user folder e.g. (\AppData\Roaming\MAXON\Maxon Cinema 4D R21\teamrender\users) to send it back to the nodes. Super inefficient and unnecessary… This cant be correct. It seems like Cinema4D or TurbulenceFX ignore the absolute path to the Cache and always pack the whole project when starting the render via teamrender…
What am i missing?!
Have a dig around here using a variety of search terms may find a solution til Jascha gets on the case https://old-forum.jawset.com/viewforum.php?f=4&sid=74acf7aedb6c067465e00439d3336af4
Unfortunately Team Render collects and distributes all assets unconditionally.
In order to work around this, i have added an option to disable cache distribution via TR entirely. The new beta build 1455 has this option in Render Settings/TurbulenceFD Renderer/Automatically Distribute Cache to Team Render Clients.
If disabled, the cache base path or alternate cache base path must be valid on the TR clients, as would be the case in your setup as i understand it.
Wow, that was fast!
I’ll try it out, many thanks!
Works like a charm! Perfect!
I take it you danced the fandango around the room !!!
Hm, after a lot of testing and first enthusiasm something is still wrong… i get absolutely random error messages on the render clients
TFD ERROR: CacheFrameGuard::load(): render server request failed for “Y:/PATH_TO_CACHE.bcf”
TFD ERROR: Failed to load frame 238 from cache “Y:/PATH_TO_CACHE” for object TurbulenceFD Container
sometimes they behave correct if i restart them, sometimes not… no way to reproduce the error in a meaningful manner…
this is absolutely random… any idea what could be the reason for this? the network ist stable, everything else in the scene is loaded correctly
Edit: Nevermind, i think this only happens, if i simulate to the local drive, then move the cache to the NAS and relink it… The workaround is to simulate directly to the mapped drive and to never touch the cache again, then it seems to work properly
What would be appreciated by many is if you could make a walk through of the steps you take to Team Render. The tricks, pitfalls and caveats.
This sounds like it might be a filesystem related issue. If TFD logs an error about not being able to load a file, it’s just forwarding errors it gets from the filesystem. When batch-copying files to a network share there may be a considerable delay between the files being written on the server and a client getting the new files listed.