SGI™ 2400 and 2800 Server Owner’s Guide Document Number 007-4192-001
CONTRIBUTORS Written by Mark Schwenden and Pablo Rozal Illustrated by Dan Young and Cheri Brown Production by Amy Swenson and Linda Rae Sande Engineering contributions by David Alexander, Ted Wong, Sam Sengupta, Bob Marinelli, Rob Bradshaw, Steve Whitney, Jim Ammon, Richard Singer, and Ben Fathi © 1999, Silicon Graphics, Inc.— All Rights Reserved The contents of this document may not be copied or duplicated in any form, in whole or in part, without the prior written permission of Silicon Graphics, Inc.
by writing to the U.S. Government Printing Office, Superintendent of Documents, Mail Stop SSOP, Washington, D.C. 20402-9328, ISBN 0-16-041736-8. VDE 0871/6.78 This equipment has been tested to and is in compliance with the Level A limits per VDE 0871.
TUV R geprufte Sicherheit NRTL/C VCCI Class 1 Statement for Japan Chinese Class A Regulatory Notice IRIX is a registered trademark and SGI, the SGI logo, Origin Vault, S2MP, IRIS InSight, and XIO are trademarks of Silicon Graphics, Inc. CrayLink is a trademark of Cray Research, Inc. VME is a trademark of Motorola. UNIX is a registered trademark in the United States and other countries, licensed exclusively through X/Open Company, Ltd.
Contents List of Figures ix List of Tables xiii About This Guide xv Finding the Information You Need xvi SGI 2400 and 2800 Server Owner’s Guide xvii IRIX Admin: Software Installation and Licensing Guide IRIS InSight Online Documentation and Help xvii Online Man Pages xvii Release Notes xviii World Wide Web Accessible Documentation xviii Conventions xix 1. 2.
Contents Major Components 21 System Controller 22 BaseIO Board 25 CPU Node Board 26 Router Boards 30 Power Distribution Unit (PDU) CrayLink Interconnect 32 PCI Carrier Assembly 34 3. Interface and Cabling Information 35 The Ethernet Interface Connection 36 Standard Serial Ports 38 Standard SCSI Connector 40 4. System Configurations 43 Overview 43 Basic System Configurations 44 Rackmount Configurations 47 Multirack Configurations 53 5.
Contents General Procedures 82 Powering Off a Rackmount System 82 Opening the Drive Door 89 Opening the Cable Cover Door 90 Specific Procedures 91 Removing a Drive Module 91 Removing the Facade 94 Removing the System Controller and CD-ROM Module XIO Board Slots 98 7. Using the System Controller 99 MMSC 99 MMSC Front Panel Display 100 MMSC Assembly 104 MSC Front Panel 109 Understanding the MSC LEDs and Switches MSC Features and Functions 113 MSC Status Messages 115 112 8.
List of Figures Figure i Figure ii Figure iii Figure iv Figure 1-1 Figure 1-2 Figure 1-3 Figure 1-4 Figure 1-5 Figure 1-6 Figure 1-7 Figure 2-1 Figure 2-2 Figure 2-3 Figure 2-4 Figure 2-5 Figure 2-6 Figure 2-7 Figure 2-8 Figure 2-9 Figure 2-10 Figure 2-11 Figure 3-1 Figure 3-2 Figure 3-3 Figure 4-1 Figure 4-2 Information Sources for the SGI 2400 and 2800 Rackmount System xvi System Owner’s Guide xvii IRIX Admin: Software Guide xvii IRIS InSight Icon xvii SGI 2400 Rackmount Server System 2 SGI 2400 Multira
List of Figures Figure 4-3 Figure 4-4 Figure 4-5 Figure 4-6 Figure 4-7 Figure 4-8 Figure 4-9 Figure 5-1 Figure 5-2 Figure 5-3 Figure 5-4 Figure 5-5 Figure 5-6 Figure 5-7 Figure 5-8 Figure 5-9 Figure 6-1 Figure 6-2 Figure 6-3 Figure 6-4 Figure 6-5 Figure 6-6 Figure 6-7 Figure 6-8 Figure 6-9 Figure 6-10 Figure 6-11 Figure 6-12 Figure 7-1 Figure 7-2 Figure 7-3 Figure 7-4 Figure 7-5 x Single Server Module in a Rack with Four Vault Drive Boxes 49 16P SGI 2400 Server 51 16P SGI 2400 Rackmount Server with Xpres
List of Figures Figure 7-6 Figure 7-7 Figure 7-8 Figure 7-9 Figure 7-10 MMSC to MMSC 10BaseT Ethernet Cabling (2 racks) 106 MMSC to 10BaseT-Hub 1BaseT Ethernet Cabling (2 racks) 107 Processor Assembly Location in Rack 108 MSC Status Panel, Serial Connector, and Switches 109 System Controller Rear Serial Connector 111 xi
List of Tables Table 2-1 Table 2-2 Table 3-1 Table 3-2 Table 4-1 Table 4-2 Table 4-3 Table 7-1 Table A-1 Table A-2 SGI 2400 and 2800 Major Components 18 BaseIO Connectors 26 Ethernet 100-BASE T Ethernet Port Pin Assignments 36 68-Pin Single-Ended, High-Density SCSI Pinouts 40 Single Module Chassis Configurations 47 Dual Module Chassis Configuration 48 Multiple-Rack Configurations 53 System Controller Messages 115 Physical and Environmental Specifications 123 Electrical and Cooling Specifications 124 xiii
About This Guide Welcome to the SGI 2400 and 2800 server systems. The SGI 2400 and 2800 Server Owner’s Guide is your complete guide to operating your rackmount system.
About This Guide Finding the Information You Need The SGI 2400 and 2800 ships with the following sources of information, as shown in Figure i.
About This Guide SGI 2400 and 2800 Server Owner’s Guide SGI 2200 Server Owner's Guide Figure ii System Owner’s Guide Refer to this manual (see Figure ii) to set up the system and install hardware options, or to find solutions when you suspect there is a hardware problem. In addition, it contains basic information on installing and removing optional software.
About This Guide For additional information about displaying reference pages using the man command, see man(1). In addition, the apropos command locates man pages based on keywords. For example, to display a list of man pages that describe disks, enter the following command at a shell prompt: apropos disk For information about setting up and using apropos, see apropos(1) and makewhatis(1M).
About This Guide Conventions The SGI 2400 and 2800 Server Owner’s Guide uses these conventions: • References to documents are in italics. • References to other chapters and sections within this guide are in quotation marks. • Names of IRIX man pages or commands that you type at the shell prompt are in italics as are IRIX filenames. • Steps to perform tasks are in numbered sentences. When a numbered step needs more explanation, the explanation follows the step.
Chapter 1 1. Introducing the SGI 2400 and 2800 Server Systems The SGI 2400 and 2800 rackmount servers provide a highly configurable system architecture that is available in a single rackmount or multirack setup. Each rackmount system consists of 2 to 16 CPUs, 64 MB to 32 GB of main memory and can provide a wide variety of I/O interfaces (see Figure 1-1). The SGI 2400 server system uses up to four racks (64 CPUs) but does not use a metarouter.
Chapter 1: Introducing the SGI 2400 and 2800 Server Systems Figure 1-1 2 SGI 2400 Rackmount Server System
System Features Figure 1-2 SGI 2400 Multirack System Configuration Example 3
Chapter 1: Introducing the SGI 2400 and 2800 Server Systems Figure 1-3 SGI 2400 Multirack Configuration (Four Racks Shown) Hardware Overview As shown in Figure 1-4, a single rackmount system can consist of up to two fully integrated and independent subsystems—Module A and Module B. Each of the modules in Figure 1-4 has a dedicated System Controller, which monitors module status.
Hardware Overview The modules communicate using the high-speed (1600 MB/sec) CrayLink Interconnect link. The CrayLink Interconnect (also known as the interconnection fabric) link consists of a set of high-speed routing switches and cabling that enables multiple connections to take place simultaneously. Using the CrayLink Interconnect, hardware resources (including main memory) can be shared and accessed by other modules in the configuration.
Chapter 1: Introducing the SGI 2400 and 2800 Server Systems Module B Fault 1.7 VDC 2.4 VDC 3.
Hardware Overview XIO 2 Node 1 Node 2 Node 3 Node 4 Node slots Block Diagram XIO 3 XIO 5 XIO 7 XIO 9 XIO 11 Router 1 XIO 4 XIO 6 XIO 8 XIO 10 XIO 12 Router 2 XIO 1 Rear Module Diagram XIO slots Node 1 External connections (to router boards in other chassis) Router 1 Module A External connections (to router boards in other chassis) Figure 1-5 XBOW 1 XIO XBOW 0 XIO XBOW 1 XIO Node 3 Node 4 Router 2 Node 1 External connections (to router boards in other chassis) XIO Node2 Modu
Chapter 1: Introducing the SGI 2400 and 2800 Server Systems Routers Module System Controller CD-ROM Node 4 Node 3 Node 2 Node 1 Xbow ASICs Hub 64-bit processors Midplane Rear of module Figure 1-6 8 I/O board A Look Inside the Module Chassis
About the SGI 2400 and 2800 About the SGI 2400 and 2800 As illustrated in Figure 1-7, the server is a number of processing modules linked together by the CrayLink Interconnect. Each processing module contains either one or two processors, a portion of main memory, a directory to maintain cache coherence, and two interfaces: one that connects to I/O devices and another that links system nodes through the CrayLink Interconnect. Cache coherence is the ability to keep data consistent throughout a system.
Chapter 1: Introducing the SGI 2400 and 2800 Server Systems Node Node board 64-bit CPU 64-bit CPU Cache Cache Directory/ Main Memory XIO Hub Router to CrayLink Interconnect Node board Node board Router Node board Router Router CrayLink Interconnect Router Node board Figure 1-7 10 Router Router Node board Router Node board Modules in an SGI 2400 or 2800 System Node board
About the SGI 2400 and 2800 The CrayLink Interconnect links modules to one another. The CrayLink Interconnect may appear to be a type of super data bus, but it differs from a bus in several important ways. A bus is a resource that can only be used by one processor at a time. The CrayLink Interconnect is a mesh of multiple, simultaneous, dynamically allocatable connections that are made from processor to processor as they are needed.
Chapter 1: Introducing the SGI 2400 and 2800 Server Systems The CrayLink Interconnect provides a minimum of two separate paths to every pair of SGI 2400 or 2800 modules. This redundancy allows the system to bypass failed routers or broken fabric links. Each fabric link is additionally protected by a CRC code and a link-level protocol, which retry any corrupted transmissions and provide fault tolerance for transient errors.
About the SGI 2400 and 2800 The SGI 2400 and 2800 memory is located in a single shared address space. Memory within this space is distributed amongst all the processors, and is accessible over the CrayLink Interconnect. I/O devices are also distributed within a shared address space; every I/O device is universally accessible throughout the system. IRIX The new 64-bit IRIX operating system is based on UNIX System V, Release 4, distributed software technology.
Chapter 2 2. Chassis Tour This chapter provides an overview of the rackmount system chassis and a description of the controls, connectors, and indicators. Overview Figure 2-1 and Figure 2-2 show the major parts of the SGI 2400 and 2800 rackmount system. These parts are briefly described in Table 2-1.
Chapter 2: Chassis Tour 2 Optional disk modules System controller/ 4 drive module door Multimodule System Controller logic module 5 CD-ROM drive 1 Module B 6 Module System Controller (MSC) 7 Multimodule System Controller (MMSC) display 8 SCSI drive box door 9 Intake baffle 3 System disk 10 Facade 11 1 Module A CrayLink Interconnect 12 Cable bail 13 Router board 14 Cable comb cover 17 Connector cover 15 Cable comb 16 Door Figure 2-1 SGI 2400 and 2800 Rackmount System (Front View) Note: In Fig
Overview 18 Multimodule system controller (MMSC) 19 Power distribution unit (PDU) 20 PDU power switch PDU off PDU on 21 Node boards 22 BaseIO 23 XIO slots 25 Main power switch 24 XIO cable guide ON Figure 2-2 OFF SGI 2400 and 2800 Rackmount Rear View Note: In Figure 2-2, the side panel is removed for clarity.
Chapter 2: Chassis Tour Table 2-1 18 SGI 2400 and 2800 Major Components Component Description 1. Module A/Module B The modules shown in Figure 2-1 are independent computing subsystems with a separate set of CPUs (central processing units) disks, System Controller, and I/O connections. There can be up to two modules in a rack and up to sixteen modules in a multirack configuration. 2.
Overview Table 2-1 (continued) SGI 2400 and 2800 Major Components Component Description 9. Intake baffle This baffle helps enable proper airflow through the rack. Note that the top of the rack has a vent as well. Airflow is generally pulled in from the top and middle of the rack and exhausted through the back and bottom of the rack. 10. Facade The removable facade covers the power supply and router boards for a module chassis. 11.
Chapter 2: Chassis Tour Table 2-1 (continued) SGI 2400 and 2800 Major Components Component Description 20. PDU switch The PDU switch is the main circuit breaker for the entire rack assembly. 21. Node board The Node board is the main processing board in the SGI 2400 or 2800 system. It contains one or two 64-bit CPUs, the hub (which provides an interface to the I/O subsystem and the CrayLink Interconnect), a portion of main memory, as well as directory memory.
Major Components Node boards 3 and 4 also have corresponding XIO slots indicated by a circle or a triangle. For example, if Node boards are installed in slots 1 and slot 3, then only the corresponding XIO slots (1 through 6) are activated.
Chapter 2: Chassis Tour System Controller There are two types of System Controllers used in the servers—a single module System Controller (MSC) and a multimodule System Controller (MMSC) and display. The MSC does not have the same functional abilities as the MMSC. For detailed information on operating the System Controller, see Chapter 7, “Using the System Controller.” MSC and Display The MSC and display (Figure 2-4) provide environmental and status monitoring for an individual system module.
Major Components Module NMI switch Module reset switch Fan hi-speed indicator LED AC OK LED DC OK LED Ambient overtemperature LED 8-digit LED display Security key switch 8-pin mini DIN diagnostic port ic Diagnost Port Standby On Diagnostic Figure 2-4 MSC and Display 23
Chapter 2: Chassis Tour Display panel Menu/Cancel Focus A c t i v i t y 1.
Major Components BaseIO Board The BaseIO board provides the basic I/O connections for the system. Figure 2-6 shows the BaseIO board and Table 2-2 describes the connector functions.
Chapter 2: Chassis Tour Table 2-2 lists a description of the connectors on the BaseIO. Table 2-2 BaseIO Connectors Connector Type Connector Description Connector Function 100-Base T 8-pin Jack 100-Mb per second Ethernet Serial 9-pin DIN RS-232 and 422 Serial SCSI 68-pin (FAST-20) Ultra SCSI (Single-ended) CPU Node Board Figure 2-7 shows the location of the CPU Node board in the chassis. The Node board is the main processing board in the SGI 2400 and 2800 systems.
Major Components Premium Directory Memory The optional premium directory DIMMs are required only for configurations with more than 16 Node boards. These directory memory DIMMs maintain cache coherence in large system configurations. Cache coherence helps provide data consistency when multiple processors need to access the same piece of memory.
Chapter 2: Chassis Tour Node 3 Node 4 Figure 2-7 28 Node 2 Node 1 Node Board Positioning in the Chassis
Major Components Bulk head LEDs Compression mounting bolt Compression mounting bolt Figure 2-8 Front View of Node Board Showing Bulkhead 29
Chapter 2: Chassis Tour Router Boards The Router board (see Figure 2-9) is a multiported, bidirectional data packet controller that can transport up to 1600 MB/sec per port (in each direction). Each system module can have one, two, or no Router boards (depending on the number of Node boards that are present). The router interfaces with the hubs on Node boards and allows the 64-bit processors on one Node board to directly access the main memory located on other Node boards.
Major Components One external port that connects only to companion rack Router board port Null Router Board Figure 2-9 Star Router Board Three external Router ports for interconnection fabric Rack Router Board System Router Boards Null Router Board The Null router board is used in deskside systems with only two Node boards. The Null router provides a low-cost method to connect two Node boards.
Chapter 2: Chassis Tour Star Router Board The Star Router board is used in deskside systems that have three or four Node boards. The Star Router is always paired with a Rack Router board for proper operation. This cost-effective router board provides connections with all the Node boards in a module but cannot be used for CrayLink Interconnect linking. The Star Router has one external connector which connects to a port on the companion CrayLink Interconnect router board through a jumper.
Major Components Crosstown cable CrayLink Interconnect and Xpress Links cable Figure 2-10 )) ) )) ) ) ) ) ) ) )) )) ))) ) ) )) )) ) )) )) ) ) ) )) ) ))) ) ) ) ))) )) ) ) ) ))))))))))))))))))))))) )))))) ) ) ) ) )))))) ))) ) )) )) )))))))))) ) ) ) ) )) ))) ) ) ) ) ) )) ) )))) CrayLink Interconnect, Xpress Link, and Crosstown Cables Cautionary Guidelines You generally should not handle the sensitive CrayLink Interconnect and Xpress link cables; they are very delicate.
Chapter 2: Chassis Tour Crosstown Cable The crosstown cables are similar to the CrayLink Interconnect and Xpress cables, so the same cautionary guidelines apply. The crosstown cables are encased in a plastic-ribbed coating and are slightly more ruggedized. The cables are used to connect to external peripherals and provide the same I/O performance as the CrayLink Interconnects and Xpress link cables.
Chapter 3 3. Interface and Cabling Information When your SGI 2400 or 2800 rackmount system is initially set up in the work area, a trained system support engineering (SSE) technician should configure and connect it. Your SGI 2400 or 2800 rackmount system is fully functional as a standalone server using Ethernet, modem, optional ATM, HIPPI, or other interconnect technologies.
Chapter 3: Interface and Cabling Information The Ethernet Interface Connection The system comes with a single 100 Base-T 8-pin Ethernet connector. Optional boards supporting additional Ethernet connectors are available. Table 3-1 shows the cable pinout assignments for the Ethernet 100-BASE T Ethernet port.
The Ethernet Interface Connection 1 2 3 4 Pin 1 Pin 2 Pin 3 Pin 4 Pin 5 Pin 6 Pin 7 Pin 8 Transmit + Transmit Receive + Reserved Reserved Receive Reserved Reserved 100 Base-T connector Figure 3-1 Standard Ethernet on SGI 2400 and 2800 Rackmount 37
Chapter 3: Interface and Cabling Information Standard Serial Ports Each SGI 2400 and 2800 rackmount system comes with two standard 9-pin serial ports. These ports can support either RS-232 or RS-422 interface devices. Figure 3-2 shows the location and pinouts for a serial port. Optional additional serial ports are also available. The RS-232 standard recommends the use of cables no longer than 50 feet (15.2 meters). This standard should also be applied to the RS-422 connections.
Standard Serial Ports Console serial port Serial port Pin 5 Ground Pin 4 Data Terminal Ready (DTR) Pin 3 Transmit Data (TD) Pin 2 Receive Data (RD) Pin 9 Ringing Indicator (RI) Pin 8 Clear to Send (CTS) Pin 7 Request to Send (RTS) Pin 6 Data Set Ready (DSR) Pin 1 Data Carrier Detect (DCD) Figure 3-2 Serial Port Location and Pinouts 39
Chapter 3: Interface and Cabling Information Standard SCSI Connector A single, external 68-pin SCSI connector is provided on the BaseIO panel (see Figure 3-3). This connector supports both Ultra SCSI and SCSI-2 devices. The connector sends single-ended SCSI signals only. Optional additional SCSI ports can be implemented using XIO option boards. The hyphen preceding a signal name indicates that the signal is low.
Standard SCSI Connector Table 3-2 (continued) 68-Pin Single-Ended, High-Density SCSI Pinouts Signal Name Pin Number Pin Number Signal Name Ground 16 50 Ground TERMPWR 17 51 TERMPWR TERMPWR 18 52 TERMPWR Reserved 19 53 Reserved Ground 20 54 Ground Ground 21 55 -ATN Ground 22 56 Ground Ground 23 57 -BSY Ground 24 58 -ACK Ground 25 59 -RST Ground 26 60 -MSG Ground 27 61 -SEL Ground 28 62 -C/D Ground 29 63 -REQ Ground 30 64 -I/O Ground 31 65
Chapter 3: Interface and Cabling Information SCSI connector (68-pin) Figure 3-3 42 Pin 1 Pin 35 Pin 34 Pin 68 68-Pin SCSI Connector
Chapter 4 4. System Configurations This chapter briefly describes some of the many configurations that are possible with the highly modular and scalable SGI 2400 and SGI 2800 server systems.
Chapter 4: System Configurations Basic System Configurations Figure 4-1 and Figure 4-2 shows various system configurations based on the number of processors in the system. These figures provide information on the type of Router board(s) that are required to support that configuration. The diagrams also show how the SGI 2400 server system expands from a basic building block module to a multimodule, multirack configuration.
Basic System Configurations Number of Node Boards Schematic Router Board Types Configuration 1 Node Board (up to 2 processors) 64-bit Processor 64-bit Processor 2 Node Boards (up to 4 processors) N N N None N NR N N R SR Null Router Board 3 Node Boards (up to 6 processors) Router Board Star Router Board IR1 Jumper IR1 Jumper N N N R SR 4 Node Boards (up to 8 processors) Router Board Star Router Board N IR1 Jumper IR1 Jumper N N N R R N N R R N N N 8 Node Boards Rackmou
Chapter 4: System Configurations Number of Node Boards Schematic Configuration 8 Node Boards (up to 16 processors) Rack System R R R R With Xpress Links R R 16 Node Boards (up to 32 processors) Multi Rack System R R R R R R With Xpress Links R R R R R R R R R R R R R R R Figure 4-2 46 R 32 Node Boards (up to 64 processors) Multi Rack System SGI 2400 Rackmount Configuration Examples
Rackmount Configurations Rackmount Configurations Figure 4-3 through Figure 4-5 show different single rackmount SGI 2400 configurations. The single rackmount systems have either one or two modules. Table 4-1 and Table 4-2 summarize the range of hardware components that can be supported by either a one or two-module configuration.
Chapter 4: System Configurations Table 4-2 Dual Module Chassis Configuration Hardware Component Number Range CPU Node boards 2 to 8 64-bit CPUs 2 to 16 Main memory per Node board 64 MB to 4 GB (using 256 MB memory DIMMs) Main memory per chassis 128 MB to 32 GB (using eight Node boards) Number of internal SCA drives 2 to 10 Number of usable XIO slots Up to 24 for a dual-module chassis Number of Origin Vaults 1 only I/O interfaces - One to two (single-ended) SCSI connector - Two or four DB
Rackmount Configurations Drive enclosure door Drive enclosures CPU module Figure 4-3 Single Server Module in a Rack with Four Vault Drive Boxes 49
Chapter 4: System Configurations Figure 4-4 shows a 16P (or 16 CPU) rackmount system attached with CrayLink Interconnect cabling. This configuration requires two Rack Router boards in each module and supports a maximum of sixteen CPUs. In this system setup, the hardware resources (such as main memory and disk storage capacity) from one module can be quickly and efficiently accessed by the other system module.
Rackmount Configurations Node boards Midplane Router boards R1 R2 R3 R4 Figure 4-4 16P SGI 2400 Server 51
Chapter 4: System Configurations Node boards Midplane Router boards R1 R2 Xpress links R3 R4 Figure 4-5 16P SGI 2400 Rackmount Server with Xpress Links Note: Xpress links can double interconnect bandwidth in single rack systems.
Multirack Configurations Multirack Configurations Sample SGI 2400 or 2800 multirack configurations are shown in Figure 4-6 through Figure 4-9. These can be large system configurations requiring a significant amount of floor space, along with special power and cooling considerations. The multirack configuration can have between 8 and 256 processors, up to 512 GB of main memory, and up to 192 XIO slots through 16 separate system modules.
Chapter 4: System Configurations Note: It is possible for a configuration to have between 4 and 32 processors. The 16P and 32P configurations are used because they represent the maximum number of processors for one- or two-rack systems. The configuration shown in Figure 4-7 uses Xpress links to increase the transmission throughput between modules and racks. The Xpress links provide additional data routing, which reduces potential transmission latency.
Multirack Configurations Node boards Router boards R1 R2 R1 R2 R3 R4 R3 R4 Xpress Links Figure 4-7 32P with Express Links 55
Chapter 4: System Configurations This configuration consists of four racks, eight system modules, and 32 Node boards. As Figure 4-8 shows, all of the available Router ports are now in use. The 64P configuration is the largest one that can be supported using the Rack Router board. To go beyond 64P, requires the use of a meta router chassis. With a meta router, configurations of up to 128P are possible (see Figure 4-9).
Multirack Configurations 32p 1 5 32p 2 3 6 7 4 8 1 5 2 3 6 7 4 8 1 2 3 4 5 6 7 8 Cray Router 32p 1 5 Figure 4-9 32p 2 6 3 7 4 8 1 5 2 6 3 7 4 8 SGI 2800 128P Configuration 57
Chapter 5 5. System Configuration and Operation This chapter describes how to configure and operate your system correctly. Warning: The rackmount system operates on 200-240 VAC. Use extreme caution when working around this voltage. Never install or remove power cords without first turning off the equipment. Caution: The rackmount system can weigh up to 750 pounds when fully configured. Use at least two people to move the system chassis and take care that the system does not tip or become unbalanced.
Chapter 5: System Configuration and Operation Connecting Your Terminal The SGI 2400 or 2800 rackmount server requires a customer-supplied American Standard Code for Information Interchange (ASCII) terminal. Note the following caution before attempting to connect your terminal. Caution: Before plugging the terminal into either a 120 VAC or a 220 VAC outlet, be sure that the electrical rating on the UL-CSA label is in either the 100 to 120 VAC or the 200 to 240 VAC range, whichever applies.
Connecting Your Terminal Console port Terminal Figure 5-1 Connecting a Terminal 61
Chapter 5: System Configuration and Operation SCSI Configuration Requirements All SGI 2400 and 2800 rackmount systems are configured with one or two internal SCSI drive bays using single-attachment connector (SCA) drives (see Figure 5-2). These drives must all be single-ended, Ultra SCSI. These Ultra SCSI drives are capable of transmitting up to 40 MB/sec.
Loading the Single Connector Assembly (SCA) Drives Into the System Loading the Single Connector Assembly (SCA) Drives Into the System The SGI 2400 and 2800 use an SCA sled assembly for mounting into the drive bay. Disk drive modules are aligned vertically at the front of the chassis, as shown in Figure 5-2. Note that the leftmost disk drive—the system drive—is oriented differently from the others. The drive bays in the modules use single-ended, ultra SCSI drives only.
Chapter 5: System Configuration and Operation Blank SCA drive Figure 5-2 64 Installing an SCA Drive Handle in closed position
Loading the Single Connector Assembly (SCA) Drives Into the System The SCSI IDs for the SCA drives are hardwired into the server’s midplane. Figure 5-3 shows the assigned addresses. The internal SCSI bus is also terminated directly on the midplane.
Chapter 5: System Configuration and Operation Loading the CD-ROM The CD-ROM installs vertically into the chassis. Follow these instructions to load the CD-ROM into the drive: 1. Press the eject button to release the CD-ROM carrier (see Figure 5-4). 2. Hold the CD-ROM diskette by the edges and place it into the holder clips as shown in Figure 5-4. 3. Insert the CD-ROM carrier into the drive.
Loading the CD-ROM Retainers Retainers Figure 5-4 Installing a CD-ROM Into an SGI 2400 or 2800 Chassis 67
Chapter 5: System Configuration and Operation Connecting Your System to an Ethernet Network Each server comes with a 100BaseT Ethernet connector on each rack. Some racks may have two Ethernet connector ports (one on each system module). Follow these instructions to connect an Ethernet drop to your system. 1. Locate the Ethernet line, then route it to the rear of the chassis. 2. Plug the cable into the applicable connector (see Figure 5-5).
Initially Powering On the System Initially Powering On the System If you are powering on the system for the first time, or if the system is completely shut down, use these procedures to power on the server. The multimodule System Controller (MMSC) display can be used to power on and power off the entire rack configuration after initial power on (see Chapter 7, “Using the System Controller”). 1. Verify that the system power switch, located in the lower rear of the system chassis, is turned off. 2.
Chapter 5: System Configuration and Operation PDU power cable plug 220 Volt power source Figure 5-6 70 Connecting the System Power Cable
Initially Powering On the System PDU power switch PDU off PDU on Figure 5-7 Turning On the PDU 71
Chapter 5: System Configuration and Operation Module power switch ON Figure 5-8 72 OFF Powering On the System (Rear View of Chassis)
Booting Your System Booting Your System Boot your system by performing the following steps: 1. Power on the system, as described in the preceding section, “Powering On the System.” 2. Go the front the chassis module and insert the system key into the key switch, located to the right of the module System Controller’s (MSC) display. Turn the key switch to the On position (see Figure 5-9). Note: The MSC begins the power-on sequence.
Chapter 5: System Configuration and Operation Module NMI switch Module reset switch Fan hi-speed indicator LED AC OK LED DC OK LED Ambient overtemperature LED 8-digit LED display Security key switch 8-pin mini DIN diagnostic port ic Diagnost Port Standby On Diagnostic Figure 5-9 74 Entry-Level System Controller Key Positions
Booting Your System The system’s progress can be monitored on the MSC’s display. As the module powering on, you should see the messages similar to the following: POWER ON VERS x.xx (PROM version number) MOD (module number, such as MOD 1, etc) Note: Do not press any of the front panel buttons while the system is booting. Pressing the buttons during this process will abort the boot arbitration sequence.
Chapter 5: System Configuration and Operation Installing the Operating System The basic IRIX operating system is factory installed on your system disk. No software installation is required. If additional software is desired, it must be downloaded either locally (using a CD-ROM drive) or remotely over the network. See the IRIX Admin: Disks and Filesystems manual for additional information about mounting and configuring drives.
Resetting Your System Resetting Your System You can reset the system using the MMSC display (see Chapter 7, “Using the System Controller”). You can also reset your individual system modules by turning the key switch on the System Controller to the Diagnostics position. Use the scroll buttons to bring up the Reset menu. Press the Menu button to activate the menu, then press the Execute button to reset the system.
Chapter 6 6. Removing and Replacing Customer-Replaceable Units This chapter describes the installation and removal procedures for customer-replaceable units (CRUs) in the SGI 2400 and 2800 rackmount systems. The CRUs are hardware components that can be safely removed by an end user without undue exposure to high electrical power potentials.
Chapter 6: Removing and Replacing Customer-Replaceable Units System disk Optional drives System controller CD-ROM Blank drive panels Facade Cable bail Figure 6-1 80 Customer-Replaceable Units (CRUs) for the SGI 2400 or 2800 Server
General Information General Information Read the following subsections for additional safety information and required tools. Safety Information Before beginning the replacement procedures, observe these precautions. Warning: This equipment uses electrical power internally that is hazardous if the equipment is improperly disassembled. Board removal and replacement should be performed only by SGI-certified personnel. Caution: Do not attempt to move CrayLink Interconnect cables.
Chapter 6: Removing and Replacing Customer-Replaceable Units General Procedures This section describes related activities that are common to most CRU installation and removal procedures for the SGI 2400 and 2800 rackmount chassis. Powering Off a Rackmount System You can either power off an individual module or power off the entire rack configuration using the multimodule System Controller (MMSC). If you power off an individual module, the other module(s) in the configuration can still operate fully.
General Procedures Security key switch Standby (Off) Figure 6-2 Turning Off the Module Controller Caution: The “Standby” mode does not turn power completely off to the module. Voltage is still present in the system. 4. Power off the module. The switch is located in the rear of the chassis, near the AC power cord receptacle (see Figure 6-3). Disconnect the system from the power source.
Chapter 6: Removing and Replacing Customer-Replaceable Units Module power switch ON Figure 6-3 84 Powering Off a Rackmount Module OFF
General Procedures Powering Off a Single Rack Configuration Use these procedures to power off a single-rack system. 1. Power off the individual system modules, see “Powering Off an Individual Module System” on page 82. 2. Power off the MMSC display (see Chapter 7, “Using the System Controller” for additional information). Display panel Menu/Cancel Focus A c t i v i t y 1.
Chapter 6: Removing and Replacing Customer-Replaceable Units Powering Off a Multi-Module Rack Configuration See Chapter 7, “Using the System Controller” then use these procedures to completely power down an entire multi-rack configuration. 1. Power down the individual system modules (see “Powering Off an Individual Module System” on page 82). 2. Power off the individual racks. It is recommended that you start from the rightmost rack and then continue with the rack to the left.
General Procedures PDU power switch PDU off PDU on Figure 6-5 Turning Off the PDU 87
Chapter 6: Removing and Replacing Customer-Replaceable Units Turn off the rack with the multi-module display LAST!!!! Figure 6-6 88 Turning Off a Multirack Configuration
General Procedures Opening the Drive Door Use these procedures to open the drive door on a rackmount system: 1. Swing open the drive door as shown in Figure 6-7. When closing the door, push it all the way in to engage the plastic tab on the bottom of the door. Note: The door should normally be in the closed position to help keep out dust and other possible contaminants from the drives and MSC.
Chapter 6: Removing and Replacing Customer-Replaceable Units Opening the Cable Cover Door The cable cover door (see Figure 6-8) provides aesthetic shielding for the CrayLink Interconnect cabling on the rackmount chassis and between side-by-side rackmount systems.
Specific Procedures Specific Procedures The following section provides instructions for replacing the customer-replaceable units (CRUs). To replace a CRU, use Figure 6-1 to identify the appropriate unit and its position in the chassis. Then proceed to the appropriate section and perform the steps. Removing a Drive Module Disk drive modules are aligned vertically at the front of the chassis. Note that the leftmost disk drive—the system drive—is oriented differently from the others as shown in Figure 6-9.
Chapter 6: Removing and Replacing Customer-Replaceable Units Optional Disk Handle in closed position Handle in open position System Disk Handle in closed position Handle in open position Figure 6-9 1. Opening the Disk Drive Unit Module Ensure that the system is powered off and unplugged. 2. To remove a disk drive module, snap the handle to the right or left (depending on the drive orientation) to the open position. The handle is centered, as shown in Figure 6-9. 3.
Specific Procedures Handle in closed position Figure 6-10 Handle in open position Removing the Drive 93
Chapter 6: Removing and Replacing Customer-Replaceable Units To insert a disk module, follow these steps: 1. Ensure that the system is powered off and unplugged. 2. If necessary, snap the handle to the open position so that it is centered, as shown in Figure 6-10. 3. If you are adding a drive, remove the drive filler plate that covers the drive slot you want to use. 4. Align the new disk module with the drive guide, as shown in Figure 6-9. 5.
Specific Procedures 9/64 Allen head screw Figure 6-11 Fau lt 1.7 VDC 2.4 VDC 3.3 VDC Fau lt Link 1.7 Stat VDC 2.4 VDC 3.
Chapter 6: Removing and Replacing Customer-Replaceable Units Removing the System Controller and CD-ROM Module The System Controller and CD-ROM drive are packaged together in one assembly. To replace either component, you must remove and insert an entire new assembly. Caution: Do not attempt to remove the MMSC and display. This procedure should only be performed by Silicon Graphics-certified personnel. 1. Power off the system (see “Powering Off a Rackmount System” on page 82). 2.
Specific Procedures Figure 6-12 Removing the System Controller and CD-ROM 97
Chapter 6: Removing and Replacing Customer-Replaceable Units XIO Board Slots Each system comes with a mimimum of 12 XIO board slots. Various types of optional interface boards are supported in the XIO slots. These may include • peripheral component interface (PCI) • high-performance parallel interface (HIPPI) • Fibre Channel • ATM There are certain installation restrictions that must be followed when XIO boards are installed or removed.
Chapter 7 7. Using the System Controller This chapter describes how to use the two system controller types in a rackmount system. • multimodule System Controller (MMSC) and display • module System Controller (MSC) and display The MMSC in a rackmount system is an intelligent communcations device that can monitor one or more system “modules” within a rack. The MSC in a deskside system or a rackmounted system “module” is less sophisticated than the MMSC.
Chapter 7: Using the System Controller The MMSC’s front display and input panel interface is only necessary in the primary rack of a fabric-interconnected multiple-rack system. A MMSC must be installed in each rack in a fabric-interconnected multiple-rack system. Figure 7-1 shows a functional block diagram of the MMSC.
MMSC Multimodule controller To other Multimodule controllers Multimodule controller display Figure 7-2 MMSC and Front Panel 101
Chapter 7: Using the System Controller The front display and input panel mounts on a cross section piece above the lower front air intake baffle (see Figure 7-3). The front display panel’s color TFT (thin film transisitor) screen measures approximately 4 inches high and 5-1/2 inches wide (10.2 cm x 14 cm).
MMSC Powering On a Rack Follow these instructions to power on an entire rack configuration through the MMSC display. 1. Select the “Focus” menu on the display using the direction keys, then press the Enter key. Make sure that the “Al” option is highlighted 2. Go to the “Action” menu on the display using the direction keys (see Figure 7-4). 3. Select the “Power Up” option and press the Enter key. Focus Action View Configure Power Up Power Down Power Cycle NMI Reset 1.
Chapter 7: Using the System Controller Resetting the Rack System Follow these instructions to reset an entire rack configuration through the MMSC display. 1. Select the “Focus” menu on the display using the direction keys, then press the Enter key. Make sure that the “All” option is highlighted 2. Go to the “Action” menu on the display using the direction keys (see Figure 7-4). 3. Select the “Reset” option and press the Enter key.
MMSC UPPER BAY CONSOLE ALTERNATE CONSOLE LOWER BAY BASE I/O TTY1 TEST Multimodule system controller (MMSC) UPPER BAY BASE I/O TTY1 Alternate console IO3 IO4 LOWER BAY IO5 IO6 IO7 IO8 IO9 IO1 0 IO1 IO11 2 Console Figure 7-5 MMSC Cabling 105
Chapter 7: Using the System Controller Multimodule System Controller Multimodule System Controller Top view Special "null," "jumper," or "patch" 10baseT cable (018-0625-001) Figure 7-6 106 MMSC to MMSC 10BaseT Ethernet Cabling (2 racks)
MMSC Multimodule System Controller 10BaseT Hub Multimodule System Controller Multimodule System Controller Top view Standard 10BaseT cable (018-0700-001) Figure 7-7 MMSC to 10BaseT-Hub 1BaseT Ethernet Cabling (2 racks) 107
Chapter 7: Using the System Controller The MMSC interfaces listed above are augmented by system console in-and-out RS-232 connectors and a modem connector. Note that when the MMSC is connected to the RS-232 interface on a system mounted in the rack, you may not use the eight-pin DIN connector on the front of a system module’s MSC.
MSC Front Panel MSC Front Panel The MSC front panel is shown in Figure 7-9.
Chapter 7: Using the System Controller The MSC provides environmental monitoring for safe operation of the deskside system. The MSC connects to the system midplane via a transition or “bridge” board and provides easy user access to switches and displays at the front of the deskside system. In the lower right section on the back of the deskside system is a DB-9 alternate console diagnostic port serial connector that is a direct mirror of the 8-pin diagnostic connector on the front panel.
MSC Front Panel System Controller serial port System Controller Serial Port (DB-9) Pin 1 Ground Pin 6 Not Used Pin 2 Data Terminal Ready (DTR) Pin 7 Request to Send (RTS) Pin 3 Transmit Data (TXD) Pin 8 Clear to Send (CTS) Pin 4 Request Data (RXD) Pin 9 Not Used Pin 5 Data Carrier Detect (DCD) Figure 7-10 System Controller Rear Serial Connector 111
Chapter 7: Using the System Controller Understanding the MSC LEDs and Switches The System Controller has one keyswitch, two push buttons, and four LED indicators. The following paragraphs provide information on the use or significance of each control or indicator. The Front Panel Keyswitch selects Standby, On, or Diagnostic status for the system. The System Reset push button initiates a system-wide reset of the deskside. The keyswitch must be in the diagnostic position to use this button.
MSC Features and Functions MSC Features and Functions The MSC has the following basic features and functions: • Issues a reset signal at power-on. • A front panel-mounted keyswitch provides a soft power-off to standby condition. • A front panel-mounted push-button system reset switch. • A front panel-mounted push-button non-maskable interrupt (NMI) switch. • Monitors ambient incoming air temperature into the system and adjusts fan speed accordingly (two speeds).
Chapter 7: Using the System Controller 114 • Provides ability to request the system serial number and configuration information via the I2C Interface. • Eight-digit alphanumeric status display. This display is updated by the System Controller or the Node cards in the system via the I2C interface. • Provides a seven-wire 9600 bps alternate console diagnostic port for off-line configuration and troubleshooting.
MSC Status Messages MSC Status Messages The MSC front panel has an eight-character LED readout that supplies information about system status. Table 7-1 gives a list of MSC messages and an explanation of what the impacts may be. Table 7-1 System Controller Messages Message Meaning of Message SYS OK The system is operating normally. R PWR UP The system is being powered on remotely via the MSC serial connection. POWER UP The system is being powered on from the front panel switch.
Chapter 7: Using the System Controller Table 7-1 (continued) 116 System Controller Messages Message Meaning of Message FAN FAIL A system fan has failed. If it is fan 1, 2, or 3, the system shuts down. A service call should be placed as soon as possible. POK FAIL A power OK failure occurred on an unidentified board.
Chapter 8 8. System Maintenance and Troubleshooting This chapter contains hardware-specific information that can be helpful if you are having trouble with your SGI 2400 or 2800 rackmount server. Maintaining Your Hardware and Software This section gives you some basic guidelines to follow to keep your hardware and software in good working order.
Chapter 8: System Maintenance and Troubleshooting Software Dos and Don’ts When your system is up and running, follow these guidelines: • Do not turn off power to a system that is currently running software. • Do not use the root account unless you are performing administrative tasks. • Make regular backups (weekly for the whole system, nightly for individual users) of all information. • Protect all accounts with a password.
Physical Inspection Checklist Physical Inspection Checklist Check every item on this list: • Make sure the terminal and main unit power switches are turned on. • If the system has power, check the System Controller display for any messages, then reset the system. Before you continue, shut down the system and turn off the power. Verify these connections: • The terminal cable is connected securely to the rear of the terminal and to the appropriate connector on the BaseIO panel.
Chapter 8: System Maintenance and Troubleshooting MSC Shutdown Under specific circumstances, the MSC may shut down the system. Usually this occurs when the operating environment becomes too warm because of fan failure, high ambient temperatures, or a combination of the two.
Recovering from a System Crash Recovering from a System Crash Your system might have crashed if it fails to boot or respond normally to input devices such as the keyboard. The most common form of system crash is terminal lockup—a situation where your system fails to accept any commands from the keyboard. Sometimes when a system crashes, data may be damaged or lost. Using the methods described in the following paragraphs, you can fix most problems that occur when a system crashes.
Appendix A A. System Specifications Table A-1 and Table A-2 provide technical specifications for the SGI server system.
Appendix A: System Specifications Table A-2 Electrical and Cooling Specifications Parameter Specification Voltage: 187-264 Volts, 1-phase Watts (from-the-wall): maximum 5750 watts Power Factor: minimum 0.98 Inrush Current: maximum 400 Frequency: Heat Output: 124 47-63 Hertz maximum 19,550 Btu/hr (1.
Index A F amber warning LED, 112 ambient incoming air, 113 apropos command, xviii fan failure detection, 112 field replaceable unit identifying, 91 field replaceable units, 79 C G commands apropos, xviii grelnotes, xviii makewhatis, xviii man, xvii relnotes, xviii critical fan, 112 D DC OK LED, 113 documentation, xvii available via the World Wide Web, xviii release notes, xviii General procedures, 82 grelnotes command, xviii H hardware maintenance of, 117, 118 Help, xvii I InSight documentation, xv
Index M maintenance, 117 makewhatis command, xviii man command, xvii messages from System Controller, 115 soft power-off, 113 soft power-off commands, 114 System Controller front panel, 109 messages, 115 serial connectors, 110 system reset switch, 113 system-wide reset, 112 N NMI switch, 112 non-critical fan, 112 non-maskable interrupt, 113 O online documentation, xvii P Powering down the system, 82 Power source, disconnecting, 83 R rear-mounted System Controller connector, 110 release notes, viewing,