Using Multiple ShotReactor Servers

It is possible to use multiple Shot Reactor servers. Shot Reactor servers can be local or remote. To enable the use of multiple Shot Reactor servers, simply add the server identification in the ShotReactor section of Lustre init.conf file (as shown below) and restart the service, using the ShotReactor executable in the Lustre home folder.

<HostName string="localhost" />
<HostName string="Server1" />
<HostName string="192.168.2.2" />
<HostName string="10.10.10.1" />

Note: The timeout token is now obsolete and can be removed from the configuration file.

When using multiple servers, the rendering tasks are divided by the amount of available servers. For exemple, for a shot that is 100 frames long, using two servers will dispacth the first half of the task to the first server (staring at frame 1) and the second half to the second server (staring at frame 51).
When working in Stereo3D mode with two servers, Left and Right track tasks are sent to different servers concurrently, so both eyes are rendered at the same time. This way, you can start playback and see the Stereo3D content.

Note: Using a single Shot Reactor with Stereo3D content renders the Left and Right media interleaved so playback can be started as soon as a few frames are rendered.
The ShotReactor status (located on top right of the screen) and the ShotReactor flag are now updated when playback starts. This can be enabled or disabled using the Update In Playback option, located in the Render/ShotReactor menu. These options are saved with the user settings on a project basis.
The ShotReactor servers list, in the Render/ShotReactor menu, shows the name of the server and its status:

You can select a Shot Reactors in the Shot Reactor list and press Online to start/stop the server. Better management of rendering tasks for shots that are already part of the ShotReactor queue. Better messaging for current rendering tasks.

In a dual GPU configuration, Shot Reactor only uses a single GPU (GPU 0).
In a shared/network storage configuration, Shot Reactor does not need to be aware of the networking infrastructure, making it possible to use Gig-E, 10 GigE or InfiniBand for connectivity between workstations, storage and ShotReactor servers. Follow the steps outlined below to share your workstation's local storage to enable the Shot Reactor servers to render your content at the right location:
  1. On the creative workstation with local attached storage, export the storage. This example assumes an IB connection and local storage. Workstation InfiniBand address (for this example): 10.10.11.70.
  2. As root, edit the etc/exports file and add: /mnt/StorageMedia *(rw,no_subtree_check,async).
  3. On each Shot Reactor server, create a mount point for the storage: mkdir /mnt/Host.
  4. As root, edit the file /etc/fstab and add: 10.10.11.70:/mnt/StorageMedia /mnt/Host nfs rw,bg,hard,intr 0 0.
  5. Restart the NFS service and mount the storage on the Shot Reactor workstations: systemctl restart nfs-server.service
mount -a.
    Note: If you have to reboot the Host workstation, make sure to unmount the Host path from the ShotReactor workstations.
  6. Create a Flame Premium or a Lustre project
Import your content, grade it, and use your Shot Reactor servers for rendering.

    Note: The more Shot Reactor servers you add to your system, the more bandwidth is required. Playback performance may be impacted if you are using many servers rendering content to your storage.