C4D Net Render Client on remote workstation

Hi Everyone,

For a few years now we have been running a robust c4d net render farm across 6 macs/pcs in our office. We need to keep increasing the amount of workstations and we'd like to use our home machines.

I've set this up and we have a PC Client remotely connected to the PC net render server, using the remote IP address. This is all great, it shows up in the net render interface and chooses the frames it's going to render.

Problem is, the remote client just wont actually render any frames! It just says 'connection to server established' and never starts to render the job. The local clients render fine as usual.

I've tried disabling the windows firewalls at client and server end, set read & write permissons for the 'administrator' folder where the results save and also opened up the necessary ports on the router. I've opened FTP Port 20 as 8080 wouldnt connect.

Can anyone possibly help me with this? Many many thanks in advance - I'm sure I'm not the only one with this issue.

Thanks everyone.

This may be a long shot, but we have had similar problems within the office. Firstly, it sounds like you have your port mapping in hand or it wouldn't connect in the first place, but if you dont then I think you need ports 8080 and 1800 going to your server.

We occasionally and randomly get the same problem happening inside and outside the office. It came down to permissions on the Mac client. Sometimes repairing or resetting the permissions on all of the C4D installation worked. It's very weird....

Steve Bell
http://www.archiform3d.com

How big is your tex folder ? How fast is your connection from server to client, how many clients have to be delivered, and how long did you wait before giving up ?

From my point of view, if scene file is correctly uploaded to your client, then tex folder should also be...

You're correct when placing scene file and tex folder to your admin folder, for netrender server to see them.
When starting render from web interface, server distributes scene file and tex folder to every single client. Normally you don't have to worry about that transfer, because it's automatically done behind your back.
What I'm saying is that if you can read in your client window "starting xxx render" then "rendering image #yyy" it does mean that transfer occurs correctly, and so port mappings are correct, even with the simpliest scene file.
With heavier scene+tex files, client will quickly say "starting xxx render", but you'll have to wait that scene and tex folder are totally transferred to client before reading "rendering image #yyy".
So depending on total files size, and bandwidth, it can take quite a long time to deliver them to every client.

Renders never starting are usually symptom of incorrect port mappings on client side. But as you said a really small scene file is effectively rendered by your client, I understand that port mappings are ok, otherwise render would not even start (meaning you wouldn't read "rendering image #yyy")

Hope this is clear to you...

Have another try with a netmonitor on client machine (I do love menumeters, that is totally free, but there is also an Apple app named Activity Monitor.app, that resides in your utility folder). At the moment you start netrender on web interface, server should send files to client(s), and so you should see incoming traffic on client side... This will just eliminates long waiting for nothing in case nothing is transfered...

Given your bandwiths, upload from server is the bottleneck.
In case mb means megabits, 0,35 mb/sec is only 0,043 megabytes, and transferring 250 megabytes would nearly take 2 hours to complete.
In case of 0,35 megabytes, 250 MB would take roughly 12 minutes to complete.

How long did you wait before giving up ?

Well, if you have port forwarding issue, you can refer to my 3 years old answer, at the beginning of this thread. Here is a copy :

If at home, you're behind a NAT, then you need a port forward rule that says "incoming TCP on 1080 goes to this particular local IP" (1080 is default port)

Next refinement is to connect several clients from one IP :
All you have to do is get a different port for each client behind the router, by specifying in the client prefs dialog other ones :1081, 1082, etc…
Of course, you also need port forwarding rules for each machine...

Let me know if something is unclear...

sorry should have read more before replying - i dont think i have put my serial in net server bit on personalise.

Awesome now it's started the job...let see if it does a frame!

Quick question regarding tex folder size...does it have to pass the whole contents of the tex to the client of the web, or just the files used at that moment in the scene?

Also does it have to pass the tex files once at the beginning or for every frame?

From my understanding, the whole tex folder is uploaded to the clients, whatever file is used or not.
Than means you have to manage more or less manually this folder. One easy way is to save project from C4D to your admin folder, once you have removed all unused textures within C4D. This way you should only copy used files.
Once scene and tex files are uploaded to your client(s), they're used until the end of the render.
But if you stop and relaunch a netrender, server will push again all files to the client(s). Sounds a bit rude, but it's a good idea anyway to avoid versioning problems in your files.

So does it render now ??