HP XC System Software Administration Guide Version 3.2

Note:
Perform all system administration from the management node that provides the appropriate
service, usually the head node.
1.1.1.2 Local Storage
The local storage for each node holds the operating system, a copy of the HP XC system software,
and temporary space that can be used by jobs. When possible, ensure that jobs that use local
storage clean up files after they are run. You might need to clean up temporary storage on local
machines if jobs do not do so adequately.
1.1.2 Services
A service is the software that runs on a node to provide a given function. In an HP XC system,
nodes are assigned services, like the following examples:
• Compute
Compute nodes run the Simple Linux Utility for Resource Management (SLURM). SLURM
enables them to launch, monitor, and control jobs.
• SLURM central control daemon service
The node that provides this service runs the daemon for the SLURM central control.
• Linux Virtual Service (LVS) Director service
The LVS Director is the node that controls the placement of the user logins on the LVS ring.
• Login
The login service is assigned to any node in the LVS ring. LVS assigns a user login to one
of these nodes whenever that user requests a login to the system through the system name
or Internet Protocol (IP) address.
A login node requires a connection to the external network.
• I/O service
An I/O node provides I/O services for the system.
The I/O nodes provide access to Storage Area Networks (SANs) connected to the node,
access to file systems that are mounted locally to the node, or both.
An I/O node should reside in the utility cabinet and connect to the root Ethernet switch.
• Network Address Translation (NAT) Server
The NAT server node provides external access to the nodes in the system that do not have
an external network connection.
The system can have multiple NAT servers configured to balance the access to the external
network.
• Management Hub Services
The node that provides management hub services is a central point for management daemons,
including those that provide system logs and performance information.
• Load Sharing Facility for High Performance Computing (LSF-HPC with SLURM) master
server
LSF-HPC with SLURM is an implementation of the Load Sharing Facility (LSF) for HP XC
systems.
The node that is assigned this service is responsible for running the LSF-HPC with SLURM
daemons.
28 HP XC Administration Environment