User`s guide

007-5474-003 1
Chapter 1
1. SGI Altix XE Cluster Quick-reference
Overview
Your SGI® Altix® XE cluster system ships with a variety of hardware and software documents
in both hard copy and soft copy formats. Hard copy documents are in the packing box and soft
copy documents are located on your system hard disk in both
/opt/sgi/Factory-Install/Docs
and
/opt/sgi/Factory-Install/CFG
Additional third-party documentation may be shipped on removable media (CD/DVD) included
with your shipment.
This document is intended as an overview of some of the common operations that system
administrators may have to perform to set-up, boot, re-configure (upgrade) or troubleshoot an SGI
Altix XE cluster.
The SGI Altix XE cluster is a set of SGI Altix 1U or 2U-high servers (compute nodes), and one
or more SGI
Altix 2U-high servers (head nodes) networked together and running parallel
programs using a message passing tool like the Message Passing Interface (MPI). Systems ordered
prior to the second quarter of 2009 generally use SGI Altix XE320 servers as compute nodes and
SGI Altix XE250 servers as administrative head nodes. Most systems ordered after the second
quarter of 2009 will use SGI Altix XE340 servers as compute nodes and XE270 servers as
administrative head nodes.
Note: Altix XE cluster configurations requiring higher levels of I/O may use XE250 or XE270
servers as compute nodes. It is possible to have an XE cluster that uses all SGI Altix XE250 or
XE270 servers, or a combination of XE250/270 and XE320/340 servers used as compute nodes.
In all of these cases the head node will be either an SGI Altix XE250 or XE270 (2U) server.
Always consult with your SGI support representative before swapping nodes between pre-existing
and newer clusters.
The XE cluster is a distributed memory system as opposed to a shared memory system like that
used in the SGI
Altix 450 or SGI Altix 4700 high-performance compute servers. Instead of
passing pointers into a shared virtual address space, parallel processes in an application pass
messages and each process has its own dedicated processor and address space. Just like a
multi-processor shared memory system, a cluster can be shared among multiple applications. For