| Sun Fire X4100 and Sun Fire X4200 Servers Product Notes |    
|
| Sun Fire X4100 and Sun Fire X4200 Servers Product Notes |
|
| C H A P T E R 1 |
|
LSI Firmware, Service Processor, and BIOS Issues |
This chapter describes the LSI firmware, Sun Integrated Lights Out Manager (ILOM) Service Processor, and BIOS issues related to the Sun Fire X4100 and Sun Fire X4200 servers. It includes these topics:
| Note - If an issue statement does not specify a particular platform, the issue applies to all platforms. |
To create a RAID volume, the firmware and BIOS must write metadata at the end of the HDD. At least 64 Mbytes of unpartitioned hard-disk space are required.
If you are using LSI firmware Version 1.08 or later, the resynchronization progress indicator might stay at 0%, even though the resynchronization is happening. The resynchronization is complete when Optimal is displayed as the status.
This is expected behavior in LSI firmware Version 1.08 or later.
When using the LSI BIOS Configuration Utility, you might see the following warning when trying to create a RAID volume by migrating existing data:
The selected disk contains partitions that may not be preserved when creating an array. This data may be lost when the array is created! If you would still like to use this disk in an array, choose the Erase Disk option on the previous menu by pressing Delete (D)
If you choose the Erase Disk option as suggested in the warning, the RAID volume will still not be created and the warning will be displayed again.
The warning occurs because the utility does not recognize the new Solaris x86 OS disk partition type.
Use the fdisk utility and choose menu item 4:
Change between Solaris and Solaris2 Partition IDs
This will change a Solaris2 ID to a Solaris1 ID (0x82).
LSI MPT BIOS version 6.02 does not provide low-level format functionality at this time.
If a hard disk is removed from a system, the service processor neither logs an event nor provides a visual alert.
This is expected behavior. The service processor does not receive events from the LSI SAS controller when a hard-disk drive is hot-plugged and visual alerts do not occur.
There are several methods you can use to reset the service processor, for example:
Any of these methods could interrupt the system or cause it to hang because of the USB plug/unplug events that are initiated between the service processor and the system.
To ensure that the service processor is reset and a USB event does not hang the system, configure the OS with minimal or no USB support.
Do not reboot the system while the service processor is resetting itself or the system could hang. Instead, change the mouse mode to the desired state before booting.
Scheduling service processor resets to occur only when the system is off or in reset will also prevent any service processor interaction with the system.
If you try to install an operating system (OS) from a Samsung SN-124 CD-ROM drive using JavaRConsole, the virtual CD-ROM drive might not be recognized.
Install the OS using the image files redirected from another device.
The following traps are currently not supported:
The following SNMP issues are known limitations:
The service processor does not support SNMP traps for the SUN-PLATFORM-MIB.
Use IPMI PET traps, a form of SNMP trap, to trap errors. You can implement them by configuring alert rules in the service processor, or by configuring PEF rules directly in IPMI. See the Integrated Lights-Out Manager Administration Guide, 819-1160, for details.
Breaks are transmitted to the system serial port only from the serial management port. The Break key does not work when you enter ssh ~B in a secure shell (SSH) session, or in the JavaRConsole.
Java Remote Console might crash if you redirect an external CD-ROM to a remote client, running RHEL4 U1 (64-bit) on both the host and the remote client.
You can redirect remote storage devices to Sun Fire X4100 or Sun Fire X4200 servers by starting the Java Remote Console from the ILOM web GUI (Remote Console -> Redirection). Because the ILOM has this capability, the operating system might display redirected USB storage devices as always being attached.
If redirection is disabled, however, you might see one of the following messages if you attempt to access those devices: Drive not ready or No media found. Starting and stopping storage redirection does not add or remove the virtual devices themselves, but affects only the media in those virtual devices.
Accessibility means removing barriers that can prevent people with disabilities from participating in substantial life activities, including the use of services, products, and information. Not only does providing access offer benefits for a wide range of users, but it is also a requirement in all current federal contracts under Section 508 of the Federal Rehabilitation Act. In the commercial sector, the Americans with Disabilities Act (ADA) calls for similar considerations when reasonably accommodating current and prospective employees.
The Sun Fire X4100 and Sun Fire X4200 servers meet Section 508 accessibility requirements. However, the following accessibility issues have been noted in the SP-firmware GUI:
alerts and confirmation boxes in the GUI have generic menu titles that do not provide enough contextual information. (6274918)
The internal software incorrectly uses the name io.f0 (Input/Output Fan 0) to refer to the rear fan tray of the Sun Fire X4200 server. The name should be FT2.
The service processor Ethernet port operates only at 100 Mbyte/sec. It does not operate at 10 Mbyte/sec.
The WebGUI might display incorrect threshold values for the temperature sensors, for example. ILOM might also display random values for non-threshold sensors.
If your password contains exactly 16 characters, you will not be able to log in to the service processor (SP) using either the ILOM web GUI or CLI.
Choose a password containing fewer than 16 characters.
When you exit a CLI session, the serial port speed is reset to 9600 bps. This might cause the serial port not to work after you exit a CLI session if the speed was set to a value other than 9600 bps.
Keep the speed of the serial management port set to 9600 bps.
While connecting to the SP CLI mode via SSH, the serial console connection might intermittently lock up when the start -script command is used to log in to the SP console.
Use the SP start console command to connect to the SP console, rather than start -script command.
If the serial console connection locks up, exit the SSH session then try again using the SP start console command.
The system might hang during POST. Often the system freezes shortly after the console displays the manufacturer’s logo. This problem is caused by a a bug in the ILOM software that supports configuring the Serial Management port. This problem can be triggered by referencing the /SP/serial part of the ILOM object tree from the ILOM command line. Avoid commands such as these:
Also avoid using the Configuration/Serial Port tab in the ILOM WebGUI.
To work around this problem, clear the internal software state that causes it. Use any of the following methods:
Sun Fire X4100 and Sun Fire 4200 DIMMs (in particular, Infineon DIMMs that use IDT registers) might sometimes have a problem going in or out of the PowerDown mode and might subsequently trigger uncorrectable ECC errors that can lead to system reboots.
Upgrade to BIOS 036 or later. BIOS 36 is included with Software Release 1.2.1. For instructions on upgrading to Software Release 1.2.1, see the Sun Fire X4100 and Sun Fire 4200Sun Fire X4100 and Sun Fire 4200 Release Notes For Software Release 1.2.1, 819-4344.
BIOS 036 disables the PowerDown mode (self-refresh/low-power mode), per AMD’s recommendation. Some DIMMs are susceptible to the noise induced when entering and exiting the PowerDown mode. With the PowerDown mode disabled, the probability of UE reboot is much reduced and the system stability increases.
The system will not boot from the internal HDD if any external storage devices are connected to a PCI card plugged in to Slot 0 only.
This occurs because the BIOS scans Slot 0, which connects to the HDDs, before scanning the embedded 1064 SAS controller.
The BIOS scans PCI devices in ascending order (from low PCI address to high PCI address). The scanning priority is:
Because of constraints in the option ROM, internal HDDs connected to the embedded LSI SAS controller might not be in the boot list if a PCI card is installed in Slot 0, which connects to the external HDDs.
Install the PCI card in Slots 1-4 instead of Slot 0. Slot 0 is the only slot that cannot be used if you want to boot from an internal hard-disk drive in the server.
If an Emulex LP10000 card is plugged in to any slot on the system and its BIOS is enabled, the system does not boot up.
To boot up the system, disable the Emulex LP10000 BIOS.
If you reset the system and interrupt BIOS power-on self-test (POST) early in the boot process, the system might lose the virtual USB devices from the boot order and stop booting from the drive. The BIOS rewrites some CMOS registers during POST and does not immediately update the CMOS checksum.
Unplug and then replug the system to cause a full hardware reset. All USB devices should reappear.
If a supported host bus adapter (HBA) card is plugged in to Slot 1 (on a Sun Fire X4100 server) or Slots 1-4 (on a Sun Fire X4200 server), the card is not detected by the BIOS during the bootup process.
This error occurs because the BIOS runs out of address space after it scans the network interface cards and SAS controller. However, the card is detected at the OS level and can still be used.
To boot the system from an external storage device, connect the device to a supported HBA controller installed in Slot 0.
After installing a supported host bus adapter (HBA) card, you might not be able to boot the system to operating system level because of changes in the drive order.
After installing or removing any supported HBA card connected to an external storage device, make sure the BIOS boot order is set correctly according to your system configuration.
With the initial version 6464 of the firmware using BIOS 22, you must manually reset the date and time and load the optimal default settings in the BIOS Configuration Utility after some system events.
The requirement to reset these items manually will be removed in the first update to the firmware.
You must use the BIOS Configuration Utility to reset the date and time and load optimal defaults after these system events:
After any of the events listed above happen, use the BIOS Configuration Utility to reset the date and time and to load the optimal defaults:
1. Enter the BIOS Configuration Utility by pressing F2 while the system is booting.
2. Change the date and time on the Main menu screen of the utility.
3. Use the arrow keys to choose Load Optimal Defaults on the Exit menu.
4. Press Enter to go to the next screen.
5. Press Enter when prompted to load the optimal defaults.
6. Choose Save Changes and Exit, and then press Enter.
7. Press Enter when prompted to save the configuration changes and exit the utility.
You might notice an increased memory latency if you upgrade the BIOS image and do not also change the optimal defaults and save the setup. This issue is not specific to an operating system.
After upgrading the BIOS using the BIOS Setup utility, follow these steps:
1. Use the arrow keys to choose Load Optimal Defaults on the Exit menu.
2. Press Enter to go to the next screen.
3. Press Enter when prompted to load the optimal defaults.
4. Choose Save Changes and Exit, and then press Enter.
5. Press Enter when prompted to save the configuration changes and exit the utility.
The DIMM fault LEDs, which are supposed to light up when a noncorrectable memory error occurs, do not work.
When the server boots and runs power-on self test (POST), the warning message, NODE-n DIMMs Manufacturer Mismatch might be displayed for each DIMM.
This warning message is seen with all supported Qimonda DIMMs and is caused by the presence of a new JEDEC manufacturer’s code. The new code is the result of Infineon spinning off its DRAM unit into a new company named Qimonda. The server BIOS has not yet been updated to recognize the new manufacturer’s code.
Verify that each pair consists of two DIMMs with the same manufacturer, size, and speed. If each pair has matching DIMMs, it is safe to ignore the warning message
If you hold down the Control-Alt-Delete keys long enough for the system to reset and re-enter BIOS POST, the BIOS will enter BIOS recovery mode. This is non-destructive unless a special BIOS recovery CD or USB-floppy is attached to the machine.
Power cycling the host by pushing the power button resolves the problem and the system returns to normal operation.
When Sun Fire X4200 servers have more than three PCI cards installed in the five PCI slots and more than three cards are PXE booted, all four of the GB Ethernet ports report the following error message:
Base-code ROM ID structure was not found
This message is displayed because of the limitation on option ROM space in the server’s BIOS.
The BIOS option ROM is 128 KB. Of these 128 KB, approximately 80 KB are used by the VGA controller, the LSI controller, and the network interface card. Approximately 48 KB remain for the Option ROM.
When more than three PCI cards PXE boot, more option ROM space is requested than is available, so the PXE base code cannot be loaded.
A maximum of three PCI cards can be PXE-booted in the server. To avoid the Base-code ROM ID structure was not found message, disable option ROM on the one or more cards that do not need option ROM to boot.
1. Enter the BIOS Setup utility by pressing the F2 key while the system is booting up and performing POST.
2. On the BIOS Main Menu screen, select the PCIPnP tab to open the PCI/PnP Settings screen.
3. Change the PCIX SLOT fields to Disabled for those PCI cards that will not be PXE booted.
4. Press and release the right arrow key until the Exit menu screen is displayed.
5. Follow the instructions on the Exit menu screen to save your changes and exit the Setup utility.
| Sun Fire X4100 and Sun Fire X4200 Servers Product Notes | 819-1162-28 |
|
Copyright © 2007, Sun Microsystems, Inc. All Rights Reserved.