IBM Power Systems 775 Manual

Contents

274 IBM Power Systems 775 for AIX and Linux HPC SolutionHardware Management ConsolesThe hardware management consoles (HMCs) are used for Service Focal Point and repair andverification procedures. During the initial installation and configuration, the HMCs areassigned the frames and CECs that they monitor for any hardware failures.Service nodeThe service nodes are LPARs within a building block, which consists of a full disk image andserves the diskless OS images for the nodes that it manages. All diskless nodes require thatthe SN supporting them is running before successfully booting. Some administrativeoperations in xCAT issued on the EMS are pushed out to the SN to perform the operations ina hierarchical manner, which is needed for system administration performance.I/O nodeThe I/O node is the LPAR with attached storage. The node contains the GPFS software thatmanages the global file system for the cluster. The I/O nodes must be operational beforecompute nodes mount the global file system.Startup assumptionsSome areas are outside of the scope of this process. To draw a boundary on what hardwareis part of the start process and what is considered a prerequisite, some assumptions aremade. For example, it is assumed that the site has power and that everything is in place tobegin the start process, including the site cooling is up and operational and all power to thedevices (switch, EMS, HMC, frames) is ready to be applied.The network switch hardware is a gray area in this process as some network switch hardwareis part of the HPC cluster and others might be outside the cluster. For this discussion, wemake the assumption that all network switches that are customer site-specific and not HPCcluster-specific are up and operational.There are some manual tasks that are involved in this process which require a manual start ofthe equipment. There must be people available to perform these tasks and they must befamiliar with the power-on controls that are needed for each task they are to perform.Examples of these controls include powering on the Ethernet network switches, shared diskfor dual EMS, HMC, frames, and so on. These manual tasks must be performed when it istime to perform the step.This process also assumes that all initial cabling and configuration (hardware and software) iscompleted before this process is started. It is also assumed that the entire system is bootedand testes were conducted to eliminate any hardware or software problems before performingthis procedure.DependenciesAs the cluster is started, it is critical that hardware or software dependencies are up andoperational before the successful completion of a hardware or software item that features thedependency. A high-level view of the dependencies helps to outline the flow of the startupprocess. This section provides an overview of the following dependencies (details aboutaccomplishing the task or verifying its completion are not provided): Ethernet switchesAt the top of the dependencies is the HPC cluster Ethernet switch hardware and anycustomer Ethernet switch hardware. These items are the first items that must be started.