What are the instructions for configuring "headless" linux servers to act as render nodes (potentially for a windows client) using blender 2.79b? Do they all have to be on the same network, or is some kind of firewall traversal provided?
top of page
bottom of page
Honestly the best way to do this is to configure each cloud node to have a separate public IP, then you will need to ensure firewall and port forwarding rules allow traffic from the internet into your nodes on ports 9000 through 9025 (these are the default ports used, but are configurable if you want to change them so your nodes use other ports).
You then start each node, start the headless server (this assumes you have crowdrender/blender installed on each node). In your client you will need to add each node and then enter its IP address when you press connect. Pressing connect will give you a list of nodes and allow you to enter the IPs of each.
Hope this helps :)
All nodes must be on the same network for sure, if you are using linux servers as render nodes only (i.e. your win machine is the client), then you will need to make sure to allow blender to listen on TCP ports 9000 through to 9025 on each linux server.
Then you can configure each node to run headless by starting blender in background mode with the following command (which you can of course script to start when you start the OS)
The command assumes you start it from the location of the blender application.
./blender -b -noaudio -P /home/user_name/.config/blender/2.79/scripts/addons/crowdrender/src/py_3_5/serv_int_start.py -- -t "server_int_proc"
Also you may wish to check what version of python you're blender 2.79b is running. If its something other than 3.5.x then you should use a difference py_3_X in the path to the service_int_start.py file.
Hope this helps :)