Chapter 6. Troubleshooting

This chapter consists of the following sections:

System Error Messages and Remedies

A system error message indicates a problem with the computer itself and normally appears during the power-on self-test (POST), before the operating system prompt appears. Table 6-1 lists the error messages alphabetically and indicates corrective actions.


Note: For instructions on using the BIOS, see “Changing BIOS Settings”.


Table 6-1. System Error Messages

Message

Action

CMOS battery bad

Replace the lithium battery or contact your service provider.

CMOS checksum error

Enter the BIOS. load the defaults, save the configuration, and exit.

CPU BIOS update code mismatch

Contact your service provider.

Diskette drive controller error or not installed floppy disk controller error

Check and connect both ends of the floppy drive cable.

Diskette drive error

Check the CMOS settings in the BIOS; check the floppy drive cable connections.

Diskette drive A type mismatch floppy drive error

Enter the BIOS and select the proper floppy drive type.

Diskette drive B type mismatch floppy drive B error

Enter the BIOS and select the proper floppy drive type.

ECC facility fail

SDRAM might be bad. Contact your service provider.

Equipment configuration error

Modify memory configuration. Contact your service provider.

Expansion ROM allocation failed

Contact your service provider.

Hard disk controller error

Enter the BIOS or check the hard disk cable connection.

IRQ setting error

Enter the BIOS to make sure that there are no IRQ device conflicts.

I/O parity error

Contact your service provider.

I/O resource conflict(s)

Check the serial and parallel ports to make sure that there are no IRQ and I/O address conflicts. 

I2C interface or device(s) error, system halt

Shut down the system and disconnect the power cable, or contact your service provider.

PS/2 keyboard error or no keyboard connected

Check the keyboard connection to the master node.

PS/2 keyboard interface error

Replace the keyboard or contact your service provider.

PS/2 pointing device error

Check the pointing device (mouse) connection to the master node.

PS/2 pointing device interface error

Enter the BIOS and check the pointing device.

Memory Error at: MMMM:SSSS:OOO (W:XXXX, R:YYYY) where: M: MB, S: segment, O: offset, X/Y: write/read pattern

Contact your service provider.

Memory resource conflict(s)

Contact your service provider.

Memory size mismatch CPU clock mismatch

Check the memory size based on the system specifications. Enter the BIOS. load the defaults, save the configuration, and exit. If the message reappears, contact your service provider.

NVRAM checksum error

Contact your service provider.

Onboard pointing device interface error

Replace the pointing device or contact your service provider.

Real time clock error

Enter the BIOS and set the time and date.

RAM parity error

Contact your service provider.

64K system management memory bad

Contact your service provider.

SMRAM not exist

Contact your service provider.

Press Esc to turn off NMI, any key to reboot

Press Esc to disregard the NMI error. Press any other key to reboot the system. If this message appears repeatedly, contact your service provider.

As a general rule, an error message of Press F1 to continue is caused by a configuration problem (see “Changing BIOS Settings”) or by a loose cable (see “Checking Cabling”). An equipment malfunction is more likely to cause a fatal error, that is, complete system failure.

If you perform the corrective steps in Table 6-1 and still receive an error message, the cause may be an equipment malfunction. If you are sure that your configuration values are correct and your battery is in good condition, the problem may be a damaged or defective chip. In either case, contact an authorized service center for assistance. 

Changing BIOS Settings

The BIOS utility allows you to view your system's configuration settings. The system is configured at the factory. You do not need to enter the BIOS when starting the computer unless you receive an error message.

The BIOS program loads configuration values into the battery-backed nonvolatile memory called CMOS RAM, a memory area that is not part of the system RAM. If you repeatedly receive error messages, the battery may be bad. In this case, the system cannot retain configuration values in CMOS. Ask a qualified technician for assistance.

To enter the BIOS, press Ctrl Alt Esc simultaneously to enter the BIOS setup menu.

Checking Cabling

On rare occasions, cables are damaged from frequent or careless uncabling and recabling operations. Make sure that the pins on the connector are not bent.

To check internal cabling, power off the node and remove the node cover by following all instructions in “Preparing a Node for Component Replacement or Addition” in Chapter 4. Check that the internal cables are securely attached to the correct connectors.

Solving Disk Problems

If you cannot access a new disk, the disk may not be properly formatted. Consult a system administration guide for disk formatting instructions; see “System Administration References” in Chapter 3.

Troubleshooting Video Problems (SGI Graphics Cluster Series 12)

If no video output is present on an attached monitor or other video peripheral, the external SGI ImageSync might be unseated or incorrect. Follow these steps:

  1. At the back of the rack, check that the SGI ImageSync cable on the master node at the bottom of the rack is plugged into the OUT port on the master node SGI ImageSync card; see Figure 6-1.

    Figure 6-1. External SGI ImageSync Cable, SGI Graphics Cluster Series 12 System

    External SGI ImageSync Cable, SGI Graphics Cluster Series 12 System

  2. Check that this cable is plugged into the IN port on the SGI ImageSync card on the next node (channel0); see Figure 6-1.

  3. Check that the next SGI ImageSync cable connects the OUT port of the channel0 node with the IN port on the SGI ImageSync card on the next node up in the rack (channel1).

  4. Continue checking the daisy-chaining for all SGI ImageSync cabling in the rack.

  5. If the problem persists, call your service provider.

Troubleshooting Node Problems

This section explains node problems and suggests solutions.

  • The pconsole program starts, claims to attach, and goes away.

    Check the network cable on the 10-Base-T or 100-Base-T Ethernet switch to ensure that the master channel is plugged in and active.

  • When a node is powered on, it opens the CD-ROM tray

    The CD-ROM drive has a broken tray switch. Replace the CD-ROM drive.

  • When a node is powered on, its screen remains blank and the splash screen is not displayed

    Press the reset switch on the node.

    If the node still maintains a blank screen, no sync is present. Power off the node, disconnect it, pull it out of the rack (see “Pulling a Node Out of the Rack” in Chapter 4), and ensure that the graphics card is properly seated in the AGP slot.


    Warning: Be sure to pull out the anti-tip shelf before you pull out a node. Failure to do so can result in the rack tipping over. Never pull out more than two nodes at a time.

    If the problem remains, call your service provider.

  • When a node is powered on, it fails to come up and the power indicator is dim.

    Power off the node, disconnect it, and pull it out of the rack (see “Pulling a Node Out of the Rack” in Chapter 4). Reseat the SGI ImageSync card.

    If the problem persists, verify that the floppy power cable is correctly connected and is not offset by one pin.

  • During use, a monitor goes into standby mode.

    Remove the monitor cable and attach it to a different system. If the monitor comes back on, reattach the monitor cable to the original node. If the monitor remains on, replace the VGA cable; it may have a broken wire.

    If the problem persists, attach a different monitor and cable to the node to verify the problem. Reboot the node to see if the problem persists.

  • When the node is powered on, no system disk is found.

    Power off the node, remove cables, pull out the anti-tip shelf, and pull out the affected node (see “Pulling a Node Out of the Rack” in Chapter 4). Reseat the SCSI cable connector onto the hard drive.

Troubleshooting Sound Card Problems (SGI Graphics Cluster Series 12 Only)

For sound card problems, consult the manual for the SoundBlaster Live! Value sound card. You can download the manual for this card free from the manufacturer's web site, http://www.soundblaster.com.

This section explains sound card problems and suggests solutions.

  • On a Linux system, the sound card does not function.

    Verify that the sound card's master and PCM channels are not muted and that gains are raised. See http://www.alsa-project.org for other troubleshooting advice.

  • On a Windows NT system, the network on the master channel stops working.

    Ping the master node and other nodes. If the master node can ping itself but not other nodes, power off the node, disconnect it, pull it out of the rack (see “Pulling a Node Out of the Rack” in Chapter 4), and remove the sound card.


    Warning: Be sure to pull out the anti-tip shelf before you pull out a node. Failure to do so can result in the rack tipping over. Never pull out more than two nodes at a time.

    Reattach cables and boot up. Verify network functionality.

    If the problem persists, reinstall Windows NT; if the problem still persists, replace the sound card.