Page 3
This equipment generates, uses, and can radiate radio frequency energy and, if not installed and used in accordance with the instruction manual, may cause harmful interference to radio communications. Operation of this equipment in a residential area is likely to cause harmful interference, in which case the user will be required to correct the interference at his own expense.
Page 8
Kernel dump device definitions......89 Using SAM to configure dump devices into the kernel ..89 Using HP-UX commands to configure dump devices into the kernel90 Runtime dump device definitions .
Preface The Operator’s Guide HP 9000 V-Class Servers documents the information necessary to operate and monitor HP V-Class servers. This book is not intended to be a tutorial but a reference for system administrators, system operators, or system managers. Notational conventions This section describes notational conventions used in this book.
Page 14
Preface Notational conventions Horizontal ellipses In command examples, horizontal ellipses (...) show repetition of the preceding items. Vertical ellipses Vertical ellipses show that lines of code have been left out of an example. indicates the keyboard keys you must Keycap Keycap press to execute the command example.
Overview The V-Class server can contain four to 16 HP PA-8200, Precision Architecture - Reduced Instruction Set Computer (PA-RISC) processors and can support memory configurations of 256 Mbytes to 16 Gbytes. This chapter provides a block diagram of the V-Class server and discusses its features and configurations.
Overview The PA-8200 processor The PA-8200 processor The V-Class server uses the Hewlett-Packard PA-8200 processor, based on the concept of Reduced Instruction Set Computers (RISC). The PA-8200 was designed according to Hewlett-Packard’s PA-RISC Architecture version 2.0 specifications. The PA-RISC architecture is presented in the PA-RISC 2.0 Architecture reference manual.
A conceptual block diagram of the node is shown in Figure 1. Centrally located in the diagram is the HP Hyperplane crossbar that is comprised of four Exemplar Routing Attachment Controllers (ERAC). The Hyperplane crossbar allows all of the processors to access all of the available memory.
Overview The node Figure 1 Functional block diagram of a V-Class system Core Logic Bus Memory Memory Proc Proc. access controller agent control Proc (EPIC) (EPAC) (EMAC) Memory Memory Proc Proc. access controller agent control Proc (EPAC) (EPIC) (EMAC) Memory Memory Proc Proc.
Overview The node Control and status registers (CSRs) System hardware is manipulated by control and status registers, CSRs, located in the processors and controllers. CSRs provide control, status, or both to the processors and other hardware in the node. Each CSR is memory mapped and is available to all processors in the system.
Overview The node The EPAC includes special hardware called the data mover for rapid message and data movement between memory within a node. This dedicated hardware greatly improves file I/O and networking over software versions. Hyperplane crossbar—routing attachment controller The Hyperplane crossbar is comprised of four ERACs that provide an interconnect for each processor and I/O device to memory.
Overview The node Figure 2 ERAC interconnection EPAC EPAC EPAC EPAC EPAC EPAC EPAC EPAC ERAC ERAC ERAC ERAC EMAC EMAC EMAC EMAC EMAC EMAC EMAC EMAC Memory access controller The EMAC controls all accesses to memory. Each EMAC controls four banks of memory, allowing up to 32 banks in an eight-EMAC node.
Overview The node A processor accesses memory by sending a request, in the form of packets, to an ERAC. The request is then forwarded to one of the EMACs. The EMAC routes requests into even and odd pending queues. Some packets not destined for memory are routed from processor to processor through the EMAC.
Overview Shared memory Shared memory V-Class servers provide exceptionally high performance using a shared- memory architecture; it allows the developer, compilers, and applications to view the system as a number of processors sharing a large physical memory and a number of high-bandwidth I/O ports. Message passing (Data Mover) hardware provides high performance for applications developed using explicit MPI messaging.
Page 25
Overview Shared memory To write data (store) into a cache line, the cache line must be “owned” exclusively by the processor. This implies that any other copies must be invalidated. Modified cache lines must be written back to memory from the cache before being overwritten.
Indicators, switches, and displays Operator panel Operator panel The operator panel is located on the top left side of the server and contains the key switch panel, CDROM drive, optional DAT tape drive, and the LCD display. Figure 3 shows the location of the operator panel and its components.
Indicators, switches, and displays Operator panel Key switch panel The key switch panel is located on the left of the operator panel, as shown in Figure 3 on page 14. The key switch panel contains a two position key switch, a DC ON LED, and a TOC (Transfer Of Control) button, as shown in Figure 4.
Indicators, switches, and displays Operator panel CDROM drive The CDROM drive is located on the left of the operator panel, as shown in Figure 3 on page 14. Figure 5 shows the CDROM drive front panel in detail. Figure 5 CDROM drive CD tray Headphone jack...
Indicators, switches, and displays Operator panel Emergency eject Insert a paper clip into the emergency eject hole to open a jammed CD tray. Chapter 2...
Indicators, switches, and displays Operator panel Optional DAT drive The DAT drive is located on the right of the operator panel, as shown in Figure 3 on page 14. The DAT drive front panel contains two indicator LEDs and an eject button, as shown in Figure 6. Figure 6 DDS-3 DAT drive front panel Tape Clean...
Indicators, switches, and displays Operator panel Tape Clean (Activity) (Attention) Meaning LED (green) LED (amber) Flashing slowly Media caution signal. Indicates that a cartridge is near the end of its life or that the heads need cleaning. Fault Flashing slowly Power-on (starts with two steady lights) Eject button Push the eject button to remove cartridges from the tape drive.
Indicators, switches, and displays System Displays System Displays The V-Class servers provide two means of displaying status and error reporting: an LCD and an Attention light bar. Figure 7 System displays LCD display Attention light bar IOLM010 9/18/97 Chapter 2...
Power-On Self Test (POST) takes about 20 seconds to start displaying output to the LCD. POST is described in the HP Diagnostics Guide: S- Class, X-Class, and V-Class Servers. The following explains the output shown in Figure 8: •...
Indicators, switches, and displays System Displays Table 3 shows the meaning of the processor status codes. Table 3 Processor status codes Code Meaning Kernel Idle FORTH (OBP) Server User Emulator Monarch Installed but not responding during POST User disabled Processor down Table 3 describes the POST status and progress code definitions.
Attention light bar The Attention light bar is located at the top left corner on the front of the HP 9000 V-Class server as shown in Figure 7 on page 21. This light bar displays system status in three ways: •...
Teststation This chapter provides information about the Hewlett-Packard workstation that serves as a teststation for the V-Class server. An ethernet bus called the test bus connects the teststation to the utilities board located in the V-Class server chassis. A teststation is required to enable service personnel to verify and troubleshoot the V-Class system.
Teststation Teststation features Teststation features The V2200 requires an HP workstation to function as the system console. There are two options: • Product number A4082A The standard teststation consists of the basic PA9000 712/60 workstation, the special hardware to connect the workstation as a console and the teststation software.
Teststation Teststation functionality Teststation functionality Several functions of the teststation include: • Running diagnostics • Updating of Utility Board firmware • Logging of environmental and system level events • Configuration of hardware and boot parameters • Booting the Operating System The teststation is connected directly to the V-Class server via a dedicated LAN connection as well as a special RS-232 serial connection.
Teststation Teststation logons Teststation logons Two UNIX user accounts are created on the teststation during the HP- UX 10.20 operating system installation process. sppuser This user is the normal logon for the teststation during system operation, verification, and troubleshooting. Default password: spp user Please note the space between spp and user.
This is the main console window (TOP RIGHT) for the V-Class server. All POST (Power-On Self-Test) status is displayed here. The user can type commands while in HP mode (boot menu) to boot and configure the node. See Chapter 4, “Firmware (OBP and PDC)” for more information about HP mode (boot menu).
Teststation Teststation logons Using the mwm Root Menu The teststation uses the Motif Window Manager (mwm) to control the X windows on your screen. The Root Menu is mwm’s main menu. The options on the Root Menu affect your entire display. From the Root Menu, you can select menu items to create new windows, initiate diagnostic tools, and perform other tasks.
Page 45
Teststation Teststation logons • shuffle down—Moves the top window in a stack of windows to the bottom. • refresh all—Refreshes your entire X display. • restart—Stops and restarts mwm. • logout—Closes all open windows and stops mwm. Chapter 3...
Teststation Using the console Using the console The console serves as a communication device between you and your V-Class server. Virtual consoles are also used to monitor specific operations, like a system software crash dump. This section familiarizes you with the console and shows you how to: •...
The Root Menu appears. Step 2. Select logout. The teststation closes all open windows, stops mwm, and returns a HP-UX login prompt. Step 3. Log into the teststation as sppuser. The new sppconsole window displays.
Teststation Using the console Accessing the console remotely Use the sppconsole command to access the console from a system other than the teststation. Using control sequences with this command allows you to watch or to assume control of the console window. Table 6 sppconsole commands Command...
Teststation Using the console Watching the console remotely Any root user can display the console via a remote login to the teststation, so it is possible to have many different processes watching the console at the same time. This is sometimes referred to as “spy mode”.
Teststation Using the console Controlling the console remotely You can perform system maintenance or diagnostics remotely by assuming control of the console from a remote terminal. Upon gaining control of the console, you have write access to that window. Only one window can be active at a time. To assume control of the console, complete the following steps: Step 1.
Teststation Using the console Changing a console’s connection Once you have started a console as a watch or a control connection, you can change the connection type with escape characters. To change a watch window to an active console window, enter: CTRL- To change an active console window to a watch window, enter: CTRL-...
Teststation Teststation file system Teststation file system The /spp directory located on the workstation’s local disk (HP-UX version 10.20) contains all the necessary files necessary for the workstation to function as the V-Class server’s teststation. Figure 10 Teststation file system /spp (spp specific daemons)
Responsible for logging events such as HPMC (High Priority Machine Checks). /spp/scripts The /spp/scripts directory contains scripts that perform a variety of functions. Dump Configuration Manager. The hard error logger, HP Machine Check (HPMC) hard_logger data collection script. The console utility. sppconsole /spp/data The /spp/data directory contains: node_0.cfg...
Teststation Teststation file system hard_hist Log of all hard failure information. Logs the output of all suspected ACIS (Application Specific Integrated Circuits). This file may be useful in troubleshooting intermittent ACIS failures. event_log Log of all event information. A read only file which captures information generated by the ccmd daemon.
Firmware (OBP and PDC) OpenBoot PROM (OBP) and SPP Processor Dependent Code (SPP_PDC) make up the firmware on HP V-Class servers that makes it possible to boot HP-UX. This chapter discusses the boot sequence and the commands available from the boot menu.
Once a machine powers on, the firmware controls the system until the operating system (OS) executes. If the system encounters an error any time during the boot process, it stops processing and goes to HP mode boot menu. See “HP mode boot menu” on page 46 for more information.
Boot sequence Figure 11 Boot process Autoboot Boot menu Enabled? displays Prompt displays: Processor is starting the autoboot process. To discontinue, press any key within 10 seconds. Continue Press any key Automatically? to display boot menu HP-UX boots Chapter 4...
Firmware (OBP and PDC) Boot process output Boot process output The following output illustrates what typically displays on the console as the system starts up: SPP2000, POST version 4.0.0.1, compiled 1997/07/31 13:20:50 Node Id: 00000000 Monarch: PB2L Probing CPUs. Completing core SRAM initialization. Initializing main memory.
Page 59
= 1/0:0.12.0 Alternate boot path = 15/3 Console path = 15/1 Keyboard path = 15/1 System is HP9000/800/V2200 series Autoboot and Autosearch flags are both OFF or we are in HP core mode. Processor is entering manual boot mode. Chapter 4...
Firmware (OBP and PDC) HP mode boot menu HP mode boot menu In some instances, the boot menu displays; otherwise the operating system boots and the system is ready for use. The boot menu displays when one of the following occurs: •...
LS [path|flash] Displays the LIF contents (boot or flash volume) of a device. OS [hpux|sppux] Displays or sets which OS is going to boot—HP- UX or SPP-UX. For V-Class, this should be set to HP-UX. PASSword Defines the password used to control access to ForthMode.
Page 62
Firmware (OBP and PDC) HP mode boot menu Command Description PAth [PRI|ALT|CON] [path] Displays or sets primary, alternate, console, and keyboard hardware paths. Keyboard path cannot be modified. PDT [CLEAR|DEBUG] Displays or clears Page Deallocation Table (PDT) information. For use by service personnel only.
Firmware (OBP and PDC) Enabling Autoboot Enabling Autoboot AUto displays or sets the Autoboot or Search flag, which sets the way a system will behave after powering on. If Autoboot is ON, the system boots automatically after reset. If AutoSearch is ON and Autoboot is OFF, the system searches for and displays all I/O devices that the system can boot from.
Firmware (OBP and PDC) Enabling Autoboot Examples This command displays the status of the Autoboot and Autosearch flags. Autoboot:ON Autosearch:ON This command displays the current setting of the au bo Autoboot flag. Autoboot:ON This command sets the Autoboot flag ON. au bo on Autoboot:ON Chapter 4...
Firmware (OBP and PDC) HElp command HElp command The help command displays help information for the specified command or redisplays the boot menu. Syntax HElp [command] Used alone, HElp displays the boot menu. Specifying command displays the syntax and description of the named command. Examples The following example illustrate use of this command: This command displays information for the auto...
Page 66
Firmware (OBP and PDC) HElp command Chapter 4...
Starting and stopping HP-UX This chapter provides information about starting and stopping HP-UX: • Starting your HP 9000 V-Class Server • Using the console • Starting HP-UX • Stopping HP-UX Chapter 5...
Allows multiple users to access the system simultaneously. See also single-user mode. run level An HP-UX mode of operation. The /etc/inittab file defines which terminals and processes are active at each run level. run-level s Restricts user input to the console.
10 seconds to interrupt the boot process and enter bootmenu commands.) • False, you must: • Start OBP at the teststation’s HP-UX prompt by entering the following command: do_reset • Start the V-Class server using default values at OBP’s default...
Starting and stopping HP-UX Starting HP-UX Starting HP-UX You must start up, or boot, HP-UX after the operating system has been completely shut down or after you have partially shut down the operating system to perform system administration tasks. This section provides a brief overview of the process;...
• Shell on the console • Background daemon processes started by /sbin/rc • Processes that the root user invokes Restarting HP-UX from single-user mode Complete the following steps to restart your system: Step 1. Select the sppconsole window on the teststation or assume control remotely.
Starting and stopping HP-UX Starting HP-UX Restarting HP-UX from multiuser mode Complete the following steps to restart your system: Step 1. Select the sppconsole window on the teststation or assume control remotely. See “Accessing the console remotely” on page 34.
Starting and stopping HP-UX Stopping HP-UX Stopping HP-UX This section provides a brief overview of the process; for complete instructions, see the Managing Systems and Workgroups. Typically, you shut down the system to: • Put it in single-user state so you can update the system or check files systems.
Starting and stopping HP-UX Stopping HP-UX Rebooting or shutting down the system To reboot or shut down your V-Class server, perform the following steps: Step 1. Select the sppconsole window on the teststation. See “Teststation sppuser windows” on page 28.
Power This chapter describes how to configure the UPS and what to do after a power failure. Tasks covered include: • Configuring a PowerTrust UPS on V-Class systems • Monitoring the PowerTrust UPS • Handling ac power failures. Chapter 6...
(1M) after notification of users). Consider decreasing the value of shutdown_timeout_mins if shutdown (1M) will take less than 5 minutes; small systems can take advantage of this. For more information on UPS on HP-UX, see the man pages on ups_mond(1M) and ups_conf(4). Chapter 6...
• Look in a log file. UPS messages are recorded in a system log file. On HP-UX, the log files is /usr/adm/syslog, a text file which can be viewed with vi or another editor. For more information, see the PowerTrust System Guide (PN 3329- 90002) that accompanied the PowerTrust UPS.
Power If ac power fails If ac power fails If the automatic shutdown feature has been enabled, follow the procedure in “ac power failure with automatic shutdown enabled”. If the automatic shutdown feature is disabled, follow the procedure in “ac power fails without automatic shutdown enabled”...
Power If ac power fails If the unit nears full battery discharge If the unit comes within approximately three (3) minutes of full battery discharge, the reboot command with the halt option will execute, as described in “If ac power does not return” on page 64. The PowerTrust UPS will issue three beeps every 10 seconds.
Power If ac power fails 4. Set the UPS/BATTERY switch to Disable. 5. Set the MAIN and BYPASS Input breakers to the Off position. 6. Set the Output breakers to the Off position. 7. If the unit is to be stored or shipped, disconnect the battery cable. If the unit nears full battery discharge If the unit comes within approximately two (2) minutes of full battery discharge, the PowerTrust UPS will issue three beeps every 10 seconds.
Recovering from failures This chapter provides detailed information on recovering from HP-UX system interruptions. Usually, the first indication of a problem is that the system does not respond to user input. This lack of response indicates either a performance problem or system interruption.
Recovering from failures Collecting information Collecting information Providing the Response Center with a complete and accurate symptom description is important in solving any problem. The V-Class server’s teststation automatically records information on environmental and system level events in several log files. See “Teststation file system” on page 38 for more information about these files.
Ctrl-C executing command. Step 3. Check another terminal to verify that the problem is not just a console hang. Step 4. Contact your CE or the HP Customer Response Center. Chapter 7...
Recovering from failures System hangs System hangs System hangs are characterized by users unable to access the system, although the LCD display and attention light may not indicate a problem exists. The system console may or may not be hung. Use the following procedure to troubleshoot a system hang: Step 1.
Page 85
TOC reset. Used to produce a crash dump Step 7. Save the core dump file and contact the HP Response Center to have the core dump file analyzed. Refer to your service contract for the phone number of your HP Response Center. See “Fast dump” on page 79 for more information.
Recovering from failures System panics System panics A system panic is the result of HP-UX encountering a condition that it is unable to respond to and halting execution. System panics are rare and are not always the result of a catastrophe.
• Data Segmentation Fault • Instruction Segmentation Fault Step 3. Categorize the panic message. The panic message describes why HP-UX panicked. Sometimes panic messages refer to internal structures of HP- UX (or its file systems) and the cause might not be obvious.
Recovering from failures System panics 2. Take the device offline. 3. Power down the device. 4. If it is a disk drive, wait for the disk to stop spinning. 5. Power up the device. 6. Place the device back online. Step 3.
Recovering from failures System panics Step 4. If the system does not reboot by itself, reboot the computer by issuing the reset command in the console window or do_reset command at the tsh-shell window. For more information about rebooting the system see “Rebooting the system”...
Recovering from failures System panics Logical Volume Manager (LVM) related problem If the size of a logical volume that contains a file system is reduced such that the logical volume is smaller than the file system within it, the file system will be corrupted.
Recovering from failures Rebooting the system Rebooting the system Once you have corrected any problem, you are ready to reset and reboot your system. Step 1. Reset the V-Class node using one of four different methods: • Power cycle the V-Class node by turning the keyswitch to the DC OFF position then back to the On position.
Page 92
Recovering from failures Rebooting the system Check applications to be sure that they are running properly and monitor the system closely for the next 24 hours. For a short while, backups may be done more frequently than normal until confidence in the system has been restored.
Hewlett-Packard do it for you. V-Class servers using HP-UX Release 11.0 or greater employ a more efficient dump mechanism than other HP servers using previous releases of HP-UX. This mechanism is called fast dump.
When you next reboot the system, a special utility copies the memory image from the dump devices to the HP-UX file system area. Once copied, the memory image can be analyzed with a debugger or saved to tape for later analysis.
Page 95
Recovering from failures Abnormal system shutdowns It is important to have sufficient space to capture the part of memory that contains the instruction or data that caused the crash. You can define more than one dump device so that if the first one fills up, the next one continues dumping until the dump is complete or no more defined space is available.
The operator is given ten seconds to override HP-UX and continue with a selective dump. The fewer pages dumped to disk (and on reboot, copy to the HP-UX file system area), the faster the system can be back up and running.
Compressed save vs. noncompressed save System dumps can be so large that they tax the HP-UX file system area. The boot time utility, savecrash, can be configured (by editing the file / etc/rc.config.d/savecrash) to compress or not compress the data...
Page 98
first have to finish moving the pages on the dedicated dump devices to the HP-UX file system area, so that you can use a utility such as tar to bundle them for shipment. To do that, use the command /usr/sbin/crashutil instead of savecrash to complete the copy.
A large amount of time and disk space is needed to dump the entire contents of memory in a system with a large memory configuration or to copy a large memory image to the HP-UX file system area during the reboot process.
Systems configured with small amounts of memory and using only the primary swap device as a dump device might not be able to preserve the dump (copy it to the HP-UX file system area) before paging activity destroys the data in the dump area. Larger memory systems are less...
88. Compressed save vs. noncompressed save Regardless of whether you choose full or selective dump, whatever is saved on the dump devices needs to be copied to the HP-UX file system area before you can use it. NOTE With HP-UX 11.0, it is possible to analyze a crash dump directly from dump devices using a debugger that supports this feature.
Abnormal system shutdowns Partial save (savecrash -p) If the system has plenty of dump device space but is limited in HP-UX file system space, consider using the -p option for the savecrash command. This option copies only those pages on dump devices that are endangered by paging activity (i.e.
Recovering from failures Abnormal system shutdowns CLASS PAGES INCLUDED IN DUMP DESCRIPTION -------- ---------- ---------------- ------------------------------------- UNUSED 2036 by default unused pages USERPG 6984 by default user process pages BCACHE 15884 by default buffer cache pages KCODE 1656 by default kernel code pages USTACK yes, by default...
Step 5. Boot the system from the new kernel file to activate the new dump device definitions. Using HP-UX commands to configure dump devices into the kernel You can also edit your system file and use the config program to build the new kernel.
Recovering from failures Abnormal system shutdowns • The logical volume cannot be used for file system storage, because the whole logical volume is used. To use logical volumes for dump devices (no matter how many logical volumes you require), include the following dump statement in the system file: dump lvol Configuring No Dump Devices—To configure a kernel with no dump...
The /etc/fstab file You can define entries in the fstab file to activate dump devices during the HP-UX initialization (boot) process or when crashconf reads the file. The format of a dump entry for /etc/fstab looks like the following: devicefile_name / dump defaults 0 0...
HP-UX encountered a condition that it could not to handle. Sometimes the cause of the crash is apparent, but many times an in-depth analysis is required. HP-UX is equipped with a dump procedure to capture the contents of memory at the time of the crash.
The dump After the operator overrides the current dump level, or the 10-second override period expires, HP-UX writes the physical memory contents to the dump devices until one of the following conditions is true: • The entire contents of memory are dumped (if a full dump was configured or requested by the operator).
What to do after the system has rebooted? After the system reboots, make sure that the physical memory image dumped to the dump devices is copied to the HP-UX file system area then you can either package and send it in for analysis or analyze it yourself using a debugger.
HP-UX file system area. If you chose to do a partial save, the only pages copied to your HP-UX file system area during the boot process are those that were on paging devices.
Crash dump format conversion If you are analyzing a crash dump on a computer running a different version of HP-UX than the V-Class server, or if you are using a debugging tool that does not recognize the specific format of the saved file, you may not be able to analyze the crash dump in its current format.
Page 112
Recovering from failures Abnormal system shutdowns Chapter 7...
Page 114
48 HP-UX processor defined status codes Tape enabling Autoboot HP-UX system interruptions memory help command HP-UX, runs on test station access role at SPP-UX startup Hyperplane crossbar banks on/off switch interleaved Open Boot PROM See OBP Message Passing Interface...
Page 115
automatic shutdown spy mode system startup enabled starting, 32 system status log, configuring restarting, 32 consolelog file monitoring sppconsole - complex console processor Processor Dependent Code sppuser tape (PDC) sppuser windows tcsh processor status codes SPP-UX tcsh shell windows prompt OBP role in startup test bus command...