| TORQUE Resource Manager |
1.2 Basic Configuration
TORQUE only requires a few steps to get the system initially configured and running. First, the server needs to know which compute nodes it can trust. Second, the server should be initialized and configured. Finally, each of the compute nodes need to know where the pbs_server is located.1.2.1 Specify Compute Nodes
In order for the pbs_server to communicate with each of the compute nodes, it needs to know which machines to contact. Each node which is to be a part of the cluster must be specified on a line in the server nodes file. This file is located at $TORQUECFG/server_priv/nodes. In most cases, it is sufficient to specify just the node name on a line as in the following example:If the compute node has multiple processors, specify the number of processors with np=<#CPUs>. For example, if node001 has 2 processors and node002 has 4, use the following:
For a full description of this file, please see Configuring the nodes File or the PBS Administrator Guide.
1.2.2 Initialize/Configure TORQUE on the Server (pbs_server)
Once installation on the TORQUE server is complete, configure the pbs_server daemon by executing the command torque.setup <USER> found packaged with the distribution source code, where <USER> is a username that will act as the TORQUE admin. (Click here for instructions to configure manually.) This script will setup a basic batch queue to get you started. If you experience problems, make sure that the most recent TORQUE executables are being executed, or that the executables are in your current PATH.Proper server configuration can be verified by following the steps listed 1.4 Testing.
1.2.3 Initialize/Configure TORQUE on Each Compute Node
If using TORQUE's self extracting packages, and using default compute node configuration, no additional steps are required and the remainder of this section can be skipped.If manual installation is occuring or more advanced compute node configuration is required, begin mom configuration by editing the $(TORQUECFG)/mom_priv/config file on each node. Recommended settings are as follows:
Since this file is identical for all compute nodes, it can be created on the head node and distributed in parallel to all systems.
See Also