InfoDoc ID   Synopsis   Date
48408   Troubleshooting Installs of Solaris [TM] x86 on the Sun [TM] LX50 System   17 Dec 2002

Status Issued

Description

This document was developed by the product team to assist in installing Solaris 8, x86 on the LX50 system. The only supported version of Solaris 8 for this platform is 2/02 or update 7 and Solaris 9 which will be re-integrated into the release cycle starting with 12/02 or update 2 as of this writing.

Problem 1: The system does not appear to be taking input from the keyboard

Solution: You are able to use the keyboard to access the "SCSI Selector Utility" in the BIOS, but the system does does not take input when it gets to the Configuration Assistant. In this case, you may be using a USB keyboard and/or mouse. USB is not currently supported. Use a PS/2 style keyboard and mouse instead. Because there is only one port for the PS/2 style keyboard and mouse, a y-splitter is included with the LX50 system. As documented in the LX50 User Guide, the keystroke to enter setup is <F2> during the initial BIOS startup.

Problem 2: After booting from CD to install Solaris, the ITU (Install Time Update) is inserted and a "short read error" is displayed on the screen

Solution: It is possible that there is a hardware fault with the floppy drive or the bus, or the media may be faulty (either a bad diskette or a mishap when putting the ITU on the diskette). Try booting from the DCA diskette. If that fails, boot from the Linux disk and see if the floppy is accessible. If the DCA diskette boots or you are able to access the floppy when booting Linux, it is likely that the media is bad.

Try re-seating the removable CD/floppy combo by making sure it is firmly seated first (push on it), then removing and reinserting if the problem persists. Be sure to power-off the system prior to removing and re-inserting the module. It is NOT hot-swappable. (ref: LX50 User Guide, P. 5-5) Also try cleaning the drive and CD. If the failure continues, replace the hardware or media as appropriate.

Problem 3: When using PXE to install the system, either a panic or "Assertion failure" error is seen.

Solution: Bug# 4725108 has been opened to cover both scenarios. The problem is being investigated, but no solution is currently available.

Boot using the Device Configuration Assistant floppy (or CD 1 of 2) and select Net or CD boot as appropriate.

Problem 4: When using DHCP on a system with multiple networking interfaces, the wrong interface answers.

Solution: This has been reported as bug# 4721862: Solaris 9 FCS with in.dhcpd configured using graphical manager, if started on all six interfaces, misbehaves; when started only on one interface, behaves fine.

Start in.dhcpd with -i < qfe0 > and it will work. The script is located at: /usr/lib/inet/in.dhcp

Problem 5: DHCP and 'sys-unconfig' fail with an "Assertion failed" error message.

Solution: Patch# 109319-29 is available to fix this problem. Be sure to follow the instruction in the README.109319-29 file.

Problem 6: After the BIOS screens, the following WARNING messages appear on the screen when rebooting the system:

WARNING: /pci@0,0/pci-ide@f,1/ide@0(ata0):

timeout: abort request, target=0 lun 0

WARNING: /pci@0,0/pci-ide@f,1/ide@0(ata0):

timeout: abort device, target=0 lun 0

WARNING: /pci@0,0/pci-ide@f,1/ide@0(ata0):

timeout: reset target, target=0 lun 0

WARNING: /pci@0,0/pci-ide@f,1/ide@0(ata0):

timeout: reset device, target=0 lun 0

WARNING: /pci@0,0/pci-ide@f,1/ide@0(ata0):

timeout: abort request, target=0 lun 0

WARNING: /pci@0,0/pci-ide@f,1/ide@0(ata0):

timeout: abort device, target=0 lun 0

WARNING: /pci@0,0/pci-ide@f,1/ide@0(ata0):

timeout: reset target, target=0 lun 0

WARNING: /pci@0,0/pci-ide@f,1/ide@0(ata0):

timeout: reset device, target=0 lun 0

WARNING: Timeout on target 0 lun 0. Initiating recovery.

WARNING: Timeout on target 0 lun 0. Initiating recovery.

WARNING: Timeout on target 6 lun 0. Initiating recovery.

WARNING: Timeout on target 6 lun 0. Initiating recovery.

The first 8 warning messages are seen because of a bug with the ATA driver (bug# 4614976). The last 4 warnings in this example are from the Intel BIOS bug which breaks ACPI (see bug# 4719576). The latter messages will be 2 timeouts for each device - in this example there are 2 devices.

The system then resets itself and this cycle repeats. If you boot kadb (by entering 'b kadb' when asked to 'Select (b)oot or (i)nterpreter: '), you will see a message similar to:

Select (b)oot or (i)nterpreter: Size: 241597 + 21688 + 78512 Bytes

/platform/i86pc/kernel/unix loaded - 0x8c000 bytes used

SunOS Release 5.8 Version Generic_108529-13 32-bit

Copyright 1983-2001 Sun Microsystems, Inc. All rights reserved.

Cannot load drivers for /pci@1,0/pci8086,340f@7,1/sd@0,0:a

(Can't load the root filesystem)

Type any key to continue

Solution: The ITU patches for ata and pcplusmp have not been applied to the installed system. You must boot the system using the Solaris Software CD (1 of 2) or the DCA diskette and apply the ITU to the realmode boot image. When booting from the DCA or the Software 1 of 2 cdrom please be sure to select the OS hard drive as the location from which to load the kernel, otherwise you will not be running from the filesystem that requires the realmode image patched and your patching efforts will fail. Once your system is up and running, follow the instructions in the ITU README which describe "Adding an ITU diskette to an Existing Solaris System".

Problem 7: When using the Solaris 8 2/02 Installation CD, reboot fails

Solution: There is a bug with the interaction of the ITU and the fdisk partition used by the installer on the Installation CD. For Solaris 8 2/02 on the LX50, CD1 or a network installation must be done -- the Installation CD is not supported.

Problem 8: After the BIOS screens, the following fewer/different WARNING messages appear on the screen when rebooting the system:

WARNING: Timeout on target 0 lun 0. Initiating recovery.

WARNING: Timeout on target 0 lun 0. Initiating recovery.

WARNING: Timeout on target 6 lun 0. Initiating recovery.

WARNING: Timeout on target 6 lun 0. Initiating recovery.

The system then resets itself and this cycle repeats. When booting kadb (by entering 'b kadb' when asked to 'Select (b)oot or (i)nterpreter: ', you will see a message similar to:

Select (b)oot or (i)nterpreter: Size: 241597 + 21688 + 78512 Bytes

/platform/i86pc/kernel/unix loaded - 0x8c000 bytes used

SunOS Release 5.8 Version Generic_108529-13 32-bit

Copyright 1983-2001 Sun Microsystems, Inc. All rights reserved.

Cannot load drivers for /pci@1,0/pci8086,340f@7,1/sd@0,0:a

(Can't load the root filesystem)

Type any key to continue

Solution: The pcplusmp patch has not been applied. One possible reason is that KU14, KU15, or KU16 have been applied to the system, overwriting the fix initially provided through the ITU. These Kernel Updates would be applied in the form of the 108529-14, 15, or 16 patches. Notes that later version of this patch (-17 and beyond) may very well cause the same problem. To fix this, follow the instructions in the previous solution, or do a 'patchadd -u' of patch #112995-01. This is a point patch and is available at http://sunsolve.sun.com/point for download. According to it's readme (README.112995-01) it is required to be installed in single-user mode.

Problem 9: Instead of booting, the system reports the following "Non-System disk" error after the BIOS screens:

Non-System disk or disk error

Replace and strike any key when ready

Solution: The ITU, or other non-bootable, diskette is in the floppy drive. Remove the diskette and reset the system.

If you are trying to use a tip line and the 'eeprom' settings are correct, there may be something wrong with the serial cable you are using. The system has two RJ-45 serial ports (one in the front and one in the back), similar to those provided with the Netra V1 and T1 servers, and only the one in the back should be used for a tip line. Either an 8-pin RJ-45-to-DB9 adapter or a cable supporting both DB-9 and RJ-45 connectors is necessary. Detailed information about the pinout for the cable can be found on page 2-28 of the "Sun Cobalt LX50 Server User Guide".

Problem 10: After the BIOS screens, the screen is blank and input is not taken from the PS/2 keyboard

Solution: Boot using the CDROM or Device Configuration Assistant diskette. Continue through the Configuration Assistant screens and choose "Disk" as the boot device when prompted. When the system comes up, use the 'eeprom' command to check the 'output-device' and 'input-device' settings. If you want to use the screen and keyboard, 'output-device' should be set to 'screen' and 'input-device' should be set to 'keyboard'. If you are trying to use a tip line, they should be set to 'ttyb'. For further information on the use of the front and rear RJ-45 serial ports see pages 1-5 and 2-27 through 2-32 of the LX50 User Guide.

Problem 11: After the system has been installed, the mouse does not appear to be recognized for input

Solution: It is possible that "USB mouse" was selected during 'kdmconfig'. Boot to single user mode with ('b -s'), and run 'sys-unconfig'. The system runs 'kdmconfig' on reboot, where they can select a new mouse device. Note that there is a space between the 'b' and the '-s' when using this option at the boot options screen during system boot.

Problem 12: LX50 with x86 boot partition hangs during boot up somewhere between the primary and secondary boot phase (the DCA is not loaded)

Solution: This has been filed as bug# 4737684 (LX50 can't boot to an x86boot partition). This bug is seen on a system installing using the Installation CD or when the installation process uses an existing x86boot partition. The solution is to install using CD 1 of 2 and repartition the entire disk.

It should also be noted that Solaris x86 may fail to boot if there have been other operating systems installed on the system or other types of partitions defined on the drive that are unfamiliar to the Solaris bootstrap program.

Problem 13: When installing Solaris, you did not preserve the service partition and would like to recover it

Solution: During the Solaris installation process, the Service Partition, along with the other factory default installation properties, will be destroyed unless you specifically preserve it. In order to recover these attributes, use the SunLinux OS Recovery CDs (3) that come with the system.

Once recovery of the factory defaults is complete, you must reinstall Solaris -- this time explicitly keeping the Service Partition intact. Instructions for doing this follow in the solution for Problem 14. Caution: All applications will also be deleted during this process so any customer unique files should be backed up first.

Problem 14: I would like to preserve the Service Partition when installing Solaris

The Select Disk screen allows you to preserve the Service Partition. This screen follows the system identification and configuration questions (such as the Device Configuration screens, kdmconfig, and hostname) and the initial installation-related questions: type of installation, geographical location, and software selection. On the Select Disk screen:

Select the disk onto which you plan to install Solaris.

The next steps vary slightly depending on the installation method being used. Follow the steps relevant to your installation method:

For text-based installation:

1. Select F4_Edit to look at the fdisk partitions on the disk.

2. Select Edit Fdisk, and then select F2_OK. You will now see a list of partitions -- one of which is labeled 'SOLARIS'. The Service Partition is partition 1 on this screen.

3. Select partition 2.

4. Select F3_Delete.

5. Select F2_OK to confirm the deletion of the partition.

6. Repeat steps 3-5 for partitions 3 and 4.

7. Select partition 2 again.

8. Select F4_Create. Use the default value (the rest of the available disk space).

9. Select F2_OK to confirm creation of the partition.

10. Select F2_OK to continue the Solaris installation.

For graphical-based installation:

1. Select Edit FDisk. The Customize Fdisk Partitions screen is displayed. The Service Partition is partition 1 on this screen.

2. Next to partition 2, select '< unused >' using the menu.

3. Select OK to confirm the label change and the deletion of the partition.

4. Repeat steps 2 and 3 for partitions 3 and 4.

5. Select 'Solaris' next to partition 2. Use the default value (the rest of the available disk space).

6. Select OK to exit the Customize Fdisk Partition screen.

7. Select Continue to continue the Solaris installation.

INTERNAL SUMMARY:

This document is largely a copy of the troubleshooting section of Solaris 8 2/02 on Sun LX50 Product Support Document released by the Product Development Team and authored by Karyn Ritter. I have added a few words here and there where it is likely customers might mis-understand or make a mistake due to unfamiliarity with the Solaris x86 install.

Authors: Karyn Ritter & Don Bessee

e-mail: karyn.ritter@sun.com & donald.bessee@sun.com

SUBMITTER: Don Bessee BUG REPORT ID: 4725108, 4721862, 4614976, 4719576, 4737684 PATCH ID: 109319-29, 112995-01 APPLIES TO: Hardware, AFO Vertical Team Docs/Install/Boot ATTACHMENTS:


Copyright (c) 1997-2003 Sun Microsystems, Inc.