-
This repository contains my scripts for managing a Flamenco 3 render farm on Linux.
-
The code assumes that both the manager and workers share a home directory, e.g. via a NAS.
-
You need to install Blender yourself, in the created "flamenco/software"-directory.
-
Please refer to the official Flamenco installation guide if you do not understand exactly what the code in this repository does.
-
This repository is not an official repository of Flamenco.
- Clone and open the repository.
git clone https://github.com/Stinktopf/renderfarm.git
- Make the setup.sh executable:
chmod 744 ./renderfarm/setup.sh
- Run the setup:
./renderfarm/setup.sh
- Run the following command on the manager node:
crontab -l > mycron; echo -e "@reboot $HOME/Rendering/init/manager.sh" >> mycron; crontab mycron
This code ensures that when a render node has been restarted, the manager process gets started again.
- Please set the <manager_ip> in reconcileWorker.sh and run the following command on every worker node :
crontab -l > mycron; echo -e "@reboot $HOME/Rendering/init/worker.sh\n*/10 * * * * $HOME/Rendering/init/reconcileWorker.sh" >> mycron; crontab mycron
This code ensures that when a render node has been restarted, the worker process gets started again. It also checks every 10 minutes if the worker process has crashed and restarts it if necessary.
- After ten minutes at most, the workers should be visible in the Flamenco web interface.
- To uninstall the whole thing, remove the created cronjobs from cron using a code editor like VIM (or remove all cronjobs by replacing the contents of reconcileWorker.sh with crontab -r):
crontab -e
And then stop the workers:
screen -X -S flamenco-worker quit
And last but not least the manager:
screen -X -S flamenco-manager quit
- If you can only communicate with the target systems through SSH, you might be able to use X11 to access the target systems; on one with X11 installed, you can use the following command:
ssh -Y -C <addr>
and start Blender or Firefox.
- It may be that your NFS is too slow to handle the traffic, this may manifest itself in an increased number of Linux kernel errors related to NFS, check with:
dmesg
- If you change the flamenco installation you are using, remember to adjust the call in the scripts as well.