Cinelerra Render Farm Print


The reprint of this tutorial is authorized only with the active link:  


When we use such programs as Cinelerra it means that we need a lot of system resources because projects are often very large.

Cinelerra's renderfarm use quite common approach for distributive calculations (cluster systems). We have one master machine and adding connections to the slaves through preferences configuration.

All nodes in the cluster should use the same filesytem. This could be done by using NFS or Samba. We'll describe the second variant.

You need to have common filesystem, so pay attention to the same paths to the project files on each cluster node (slave machines and master machine), also we need to use the same system users to avoid problems with access rights.

Let's describe an example configuration. We have a master machine with cinelerra running and project work directory is /home/username/cinelerra/project. Next we need to share this folder between other computers in your network. This could be done easily by running nautilus and right-click on the /home/username/cinelerra/project folder and share folder. Then we need to create samba user similar to your username. This can be done by using the following command (in command line, for e.g. in gnome-terminal):
#sudo smbpasswd -a username

Then type user's password.

On cluster nodes (slave machines) you'll need to have same user accounts. Create same project folder as on master machine - /home/username/cinelerra/project. And mount remote folder there by the following command:
#sudo mount -t smbfs -o username=username,password=yourpassword //<mastercomputername>/<workdirectory> /home/username/cinelerra/project

Once this is done and you have write access to the /home/username/cinelerra/project folder you may start cinelerra as render slaves by following command:
#sudo cinelerra -d

By default cinelerra uses port 400. For test purposes it is a good idea to start cinelerra render slave in foreground:
#sudo cinelerra -f

Next we need to enable renderfarm on master machine and add render nodes. Go to the Settings -> Preferences -> Performance tab, select checkbox Use render farm. Add slave nodes (use IP adresses of computers). In "Total Jobs to create" inputbox fill in 3 times the number of rendering computers, the master included (not necessary, just recommendation).

Thats all, now you may start rendering using just configured render farm.