Setup guide

If your network supports jumbo frame switching, test if jumbo frames can be sent between the workstations
and render nodes:
1 On a workstation or render node, open a terminal and run ping using the -s option to set the packet
size used for network communications. Type: ping -s 50000 <hostname> where <hostname> is the
hostname or IP address of the workstation or render node you are trying to reach.
2 If ping fails, ensure basic network connectivity between the same two network components: ping
<hostname> where <hostname> is the hostname or IP address of the workstation or render node you are
trying to reach. If step 3 works while step 2 failed, recheck the configuration for your network adapter
and/or switch to ensure:
■ Jumbo frames are supported by both network components.
■ Both the network adapter and switch are properly configured for jumbo frame support.
The network infrastructure in your facility may not support jumbo frame switching. Consult your
network administrator if you are unsure if your network supports jumbo frames.
Check mount points on the background processing network
Certain directories on Creative Finishing workstations, such as /usr/discreet/clip and
/usr/discreet/project, must be accessible as NFS mount points to render nodes, so that material for
processing jobs can be retrieved by render nodes.
The directories shared by each Creative Finishing workstation must be mounted on the
/hosts/<workstation_hostname> directory on render nodes, where <workstation_hostname> is the host
name of the workstation submitting jobs to Burn. Normallly, these directories are mounted automatically
when you install the Creative Finishing application on the workstations, and the Burn software on the
render nodes.
Perform the following procedure to check these mount points and ensure that render nodes can access
material.
To check that mount points are accessible to the background processing network:
1 Log in as root to a render node, and open a terminal, if necessary. Type: cd
/hosts/<hostname>/usr/discreet/clip where <hostname> is the name of the workstation submitting
jobs to Burn.
2 If the previous step fails, configure the amd automounter service on each render node, and make sure
the amd and NFS services are set to run automatically on each render node and workstation. See below.
3 Try again to connect to the usr/discreet/clip directory of the workstation that is submitting jobs to
the background processing network, and create an empty file by typing:
<workstation_hostname>/usr/discreet/clip/export_test where <workstation_hostname> is the
hostname or IP address of the workstation. If the file cannot be created, permission to write to the
required directories may not be assigned to the render nodes.
4 Check the permissions assigned to the file you created in the previous step. Type: ls -al
/hosts/<workstation_hostname>/usr/discreet/clip where <workstation_hostname> is the hostname
or IP address of the workstation. File system details for the export_test file you created appear.
5 Ensure that the owner of the file is root and the group ID is sys..
Configure the amd automounter on render nodes
The amd automounter is automatically installed on each render node by the custom Linux kickstart file.
Perform the following tasks to modify the amd configuration file to redirect the mount point to /hosts
instead of /net. :
1 Log in as root to the render node. Stop the amd automounter daemon: /etc/init.d/amd stop
Burn | 127