Sun Fire 880 Server Product Notes |
The official released name for this system is Sun Fire V880. Any reference to Sun Fire 880 on the product, in the system documentation, or in software refers to the Sun Fire V880. Both names can be considered equivalent and are interchangeable.
This section provides a summarized revision history of these Product Notes, listed according to the document revision number and release date. All the updates made in previous revisions are incorporated into the latest revision; that is, any updates made in the -13 revision will carry forward through any subsequent revisions. The first published version of this document is 806-6598-12, September 2001.
|
|
Revisions Made in 806-6593-17, April 2002 (806-6598-16 is not used) |
---|
|
|
|
|
|
|
|
|
|
|
A complete set of online documentation supporting Sun Fire 880 servers is available at the following URLs:
http://www.sun.com/products-n-solutions/hardware/docs
Check these sites periodically for the latest revisions of Sun Fire 880 product documentation, including the latest version of these Product Notes.
The following table lists the minimum Solaris operating environment and OpenBoot PROM firmware requirements for Sun Fire 880 systems, according to the speed of the CPU/Memory boards installed on a Sun Fire 880 server.
Note - If you are upgrading a Sun Fire 880 server using 750 MHz CPU/Memory boards to 900 MHz, 1050 MHz, or 1200 MHz CPU Memory boards, you must update the system OpenBoot firmware with patch 112186-09 or later before you install the new CPU/Memory boards. For more information about the patch, including installation instructions, read the patch description. For information about obtaining the patch, see Required And Recommended Software Patches. |
To verify that the proper version of the operating environment is installed on the Sun Fire 880 server, examine the /etc/release file. This file should contain the text "Solaris X X/XX" or identify a subsequent compatible Solaris release.
To determine the current OpenBoot firmware version on a system, use one of the following methods.
While the Solaris operating environment is running, type the following command:
# /usr/sbin/prtconf -V |
From the ok prompt, type the following command:
ok .version |
For more information about obtaining and installing OpenBoot firmware, see Required And Recommended Software Patches.
The following table lists the CPU/Memory board part numbers for the Sun Fire 880 server.
A Sun Fire 880 server must use CPU/Memory boards of the same speed within a system. For example, you can not run 750 MHz CPU/Memory boards and 900 MHz CPU/Memory boards installed in the same Sun Fire 880 system.
It is required that you populate each DIMM slot on a 1050 MHz and 1200 MHz CPU/Memory board in a Sun Fire 880 server. For lower speed CPU/Memory boards, it is recommended that you populate each DIMM slot. For more information about DIMMs, including information about memory interleaving, see the Sun Fire 880 Server Owner's Manual.
1GByte memory modules are not supported in 750 MHz CPU/Memory board modules
This section lists software patches for the Sun Fire 880 server. You can obtain these patches from your authorized Sun service provider or by downloading them from the SunSolve OnlineSM web site at the following URL:
The patches in these Product Notes are listed by the Solaris operating system version you may have installed on your system. For the most current list of Sun Fire 880 patches, search for the Info Doc "Highly Recommended Patches Sun Fire 880" on the SunSolve Online web site. For patch installation instructions, see the README file that accompanies each patch.
Hardware/DPM: Sun Fire V880 Fibre-Channel Backplane Firmware (9228) |
|
* Patches 111847, 113039, 113040, 113041, 113042, 113043 and 113044, require the package SUNWsan (San Foundation Kit). The SUNWsan package is available via the Sun Download Center at the following URL:
http://www.sun.com/storage/san/
From that site, download the latest SAN release Software/Firmware upgrade.
* Patches 110849, 111095, 111096, 111097, 111412, 111413 and 111846 require the package SUNWsan (San Foundation Kit). The SUNWsan package is available via the Sun Download Center at the following URL:
http://www.sun.com/storage/san/
From that site, download the latest SAN release Software/Firmware upgrade.
This patch is not OS-dependant. Included in this firmware upgrade are enhanced OpenBoot tools for service providers. The patch is required for Sun Fire 880 systems using 900 MHz, 1050 MHz, and 1200 MHz CPU/Memory boards. The patch is recommended for Sun Fire 880 systems that have installed versions of OBP earlier than 4.5.6. For specific CPU/Memory board OpenBoot firmware requirements, see Software and OpenBoot Firmware Requirements. For more information about the patch, including installation instructions, read the patch description.
For a PCI card to be successfully detached from a running operating environment, each device on the card must have a detach-safe driver. A detach-safe driver enables a single instance of a driver to be closed while other instances are allowed to remain open to service similar devices used elsewhere in the system. To be considered detach-safe, a driver must be able to perform a basic Device Driver Interface/Device Kernel Interface (DDI/DKI) function called DDI_DETACH. Any driver that does not support the DDI_DETACH function is considered detach-unsafe.
Sun Microsystems offers a variety of hot-pluggable PCI cards that use detach-safe device drivers. The following table lists the PCI cards that have been tested and verified as being detach-safe as of the publication date of this document.
Note - *The Sun StorEdge PCI Dual Fibre Channel Network Adapter in the Sun Fire 880 systems requires the Solaris 8 10/01 or 8 02/02 operating environment or a compatible Solaris version. |
Note - Always wait for a PCI hot-plug operation to complete before initiating a new operation. |
For more information about Sun Fire 880 PCI hot-plug operations, refer to the Sun Fire 880 Dynamic Reconfiguration Guide (816-1465-10). This online document is available on the Solaris on Sun Hardware AnswerBook, which is provided on the Computer Systems Supplement CD for your specific Solaris release.
For an updated list of Sun PCI cards that support PCI hot-plug operations, see the latest revision of these Product Notes, available at the following URL:
http://www.sun.com/products-n-solutions/hardware/docs
In order to ensure robust system operation, it is extremely important to ensure that any PCI cards and associated drivers installed in a Sun Fire 880 system have been qualified by Sun for use on the platform. It is possible for interactions to occur between cards and drivers on a specific bus that can lead to potential system panics or other negative outcomes if the card/driver solution has not been qualified. For an updated list of qualified PCI cards and configurations for the Sun Fire 880 system, contact your Sun authorized sales representative or service provider. For additional information, refer to the following URL:
The Sun Fire 880 server uses OpenBoot 4.x system firmware. Instructions for using the firmware are provided in the OpenBoot 4.x Command Reference Manual, an online version of which is included with the OpenBoot Collection AnswerBook that ships with Solaris software.
All FC-AL disk drives installed in a Sun Fire 880 system must meet the minimum firmware revision levels for Sun Fire 880 systems. Incorrect firmware can cause a variety of system problems that are often difficult to diagnose. To determine a disk drive's firmware revision level, use the inquiry feature of the Solaris format(1M) utility. The following table shows the minimum firmware revision levels for Sun Fire 880 disk drives as of the publication date of this document.
Note - Drive capacities lower than 36 Gbytes are not supported on the Sun Fire 880. |
For the most current list of Sun Fire 880 supported disk drives, see the latest revision of these Product Notes, available at the following URL:
http://www.sun.com/products-n-solutions/hardware/docs
A backup image of the Sun Fire 880 FC-AL backplane firmware is provided on the Computer Systems Supplement CD for your specific Solaris release. In the unlikely event that the firmware on a Sun Fire 880 FC-AL backplane becomes corrupted, you can use the backup image to flash update the backplane with its original firmware. The flash update procedure is performed with the Solaris luxadm utility and is described below.
Before you can flash update the backplane, the firmware image must be copied from the Supplement CD to the Sun Fire 880 system disk. If you use Solaris Web Start to install the Supplement CD software, the Sun Fire 880 FC-AL Backplane Firmware is not included in the default installation. To install the backup firmware on the system disk, choose the Custom Install option and select the Sun Fire 880 FC-AL Backplane Firmware. For more information, see "Installing Supplement CD Software" in the Solaris on Sun Hardware Platform Guide.
Whether installed from the Supplement CD or downloaded from the SunSolve web site, the firmware image is installed in the following default location on the Sun Fire 880 system disk:
/usr/platform/SUNW,Sun-Fire-880/lib/images/int_fcbpl_fw
Once the image is installed in this location, perform the flash update procedure as follows.
1. As superuser, type the following command to bring the system to single-user mode:
# init s |
2. Place the security keyswitch in the Normal position.
The Locked keyswitch position prevents unauthorized programming of the system flash PROMs.
3. Type the following luxadm subcommand to begin the flash update process:
# luxadm download -f firmware_path enclosure_name |
Note - For more information about the luxadm utility, see Platform Notes: Using luxadm Software, part of the Solaris on Sun Hardware AnswerBook2 set on the Supplement CD. |
4. When the superuser prompt reappears, wait at least 15 more minutes for the flash process to complete.
This minimum wait time is required for the flash update process to propagate the firmware code to all SSC100 controllers in the system.
5. To verify that the flash update process has successfully completed, type the following luxadm subcommand:
# luxadm display enclosure_name |
Where enclosure_name is the enclosure name assigned to the Sun Fire 880 internal storage array.
The command output shows the status of each SSC100 in the system. The following is an excerpt of sample output for a dual-backplane system.
SSC100's - 0=Base Bkpln, 1=Base LoopB, 2=Exp Bkpln, 3=Exp LoopB SSC100 #0: O.K.(9228/ 3A20) SSC100 #1: O.K.(9228/ 3A20) SSC100 #2: O.K.(9228/ 3A20) SSC100 #3: O.K.(9228/ 3A20) |
a. Verify that each SSC100 displays an "O.K." status and that each displays the same firmware version in parentheses.
If so, the flash update process has successfully completed.
b. Otherwise, wait another two minutes or so and repeat this step.
6. Once the flash update process is complete, restore the system to multiuser mode using the init command.
# init 3 |
7. Place the security keyswitch in the Locked position.
The locked position is the recommended setting for normal day-to-day operations.
The system can now resume normal operation.
In the rare event that the system console hangs or appears to be in a reset loop, use the following procedure to recover from this situation. For more information, see "About OpenBoot Emergency Procedures" in the Sun Fire 880 Server Owner's Guide and "How to use POST Diagnostics" in the Sun Fire 880 Server Service Manual. For more information about Solaris related troubleshooting, see "Troubleshooting Software Problems" in the Solaris System Administration Guide.
1. Ensure that the system is hanging.
a. Determine if any network activity is functional via the ping command and if any existing logins from other users are active or responding.
If another login is active, use it to review the contents of /var/adm/messages for any indications of the system problem.
b. Determine if a console login session can be established through a ttya connection.
If a working console connection can be established, the problem may not be a true hang but instead a network related problem. For suspected network problems, attempt to ping, rlogin, or telnet to another system that is on the same sub-network, hub, or router that the system is on. If NFS services are served by the affected system, determine if NFS activity is present on other systems.
2. Place the system security keyswitch in the Diagnostics position.
Check your console to see if the system notes the change in the keyswitch position.
3. If there are no responding login sessions, record the state of the system LEDs.
The system LEDs may indicate a hardware failure in the system. Refer to your
Sun Fire 880 Server Owner's Guide for more information on system LEDs.
4. Attempt to bring the system to the ok prompt by issuing the Stop-A command from your keyboard.
The Stop-A command attempts to bring a system with a standard or USB keyboard to the ok prompt. If you have a TTYA terminal attached to the system, generate a break sequence.
For more information about system keyboards, see "About OpenBoot Emergency Procedures" in the Sun Fire 880 Server Owner's Guide.
5. If you are using RSC 2.2, check the RSC logs for information on the state of the system.
For further information about RSC, see the Sun Remote System Control (RSC) User's Guide.
6. If you are using RSC 2.2 and were not able to bring the system to the ok prompt in Step 4, issue the XIR command from RSC.
7. If the system is at ok prompt, do the following.
a. Issue the printenv command to display the OpenBoot configuration variables.
For more information about OpenBoot configuration variables, see "About POST Diagnostics" in the Sun Fire 880 Server Service Manual.
b. Set the diag-switch variable to true and the diag-level variable to max.
c. Issue the sync command to obtain a crash (core) dump file.
Saved core dump files provide invaluable information to your support provider to aid in diagnosing any system problems. For further information about core dump files, see "Managing System Crash Information" in the Solaris System Administration Guide.
The system will reboot automatically after issuing the sync command, provided that the OpenBoot configuration variable is set to autoboot (the default value).
8. If you were not able to bring the system to the ok prompt, ensure that the security keyswitch is in the Diagnostics position.
This forces the system to run POST and OpenBoot diagnostics during system startup.
a. Press the system Power button for five seconds.
This causes an immediate hardware shutdown.
b. Wait at least 30 seconds; then power on the system by pressing the system Power button.
9. Use the POST and OpenBoot diagnostics tools to diagnose system problems.
When the system initiates the startup sequence, it will run POST and OpenBoot diagnostics during system startup. For more information about these tools, see "Diagnostics and Troubleshooting" in the Sun Fire 880 Server Service Manual.
10. If the system is able to reboot, review the contents of /var/adm/messages for more information about the system's state. Look for the following information.
1. Any large gaps in the time stamp of Solaris or application messages.
2. Warning messages about any hardware or software components.
3. Information from last root logins to determine if any system administrators can add any comments about the system state at the time of the hang.
Currently, the only supported Sun racks for the Sun Fire 880 server are the StorEdge Expansion Rack, Sun part number SG-XARY030A, the Sun Fire Expansion Cabinet, Sun part number SF-XCAB, and the Sun Rack 900.
To rackmount a Sun Fire 880 server in the Sun Rack 900, use the X9638A rackmount kit.
To rackmount a Sun Fire 880 server in the Sun StorEdge Expansion rack, (SG-XARY030A) and the Sun Fire Expansion cabinet (SF-XCAB), use the rackmount kit X9628A.
An optional door, Sun part number X9818A, is available for the StorEdge Expansion Rack. This door is required for EMI purposes if you are rackmounting 2 Sun Fire 880s in this rack.
Note - You can only install one Sun Fire 880 server into the Sun Fire Expansion Cabinet (SF-XCAB). |
The four lifting handles included in the rackmounting kit are not designed to support the weight of a fully populated system.
The information about the Sun Fire 880 server independent memory subsystem in the Sun Fire 880 Server Owner's Guide and Sun Fire 880 Server Service Manual is misleading. In the event of a CPU failure, the DIMM groups associated with both CPUs in the failed CPU's CPU/Memory board become unavailable to all other CPUs in the system.
The procedure "How to Remove a Motherboard Fan Tray" in the Sun Fire 880 Server Service Manual is incorrect. On systems configured with the redundant cooling option, removing motherboard fan tray 5 (the primary motherboard fan tray) requires that you disconnect the cable for motherboard fan tray 6.
The procedure should include the following steps after Step 2 in the manual, in this order:
3. Loosen the captive screw on the fan tray you are going to remove.
4. Disconnect the fan tray cable from the fan tray you are going to remove.
5. If you are removing fan tray 5 and fan tray 6 is installed, disconnect the cable to fan tray 6 and drape it out of the way so that you can remove fan tray 5.
6. Slide the fan tray you are removing out of the system.
7. If you disconnected the cable to fan tray 6 to remove fan tray 5, reconnect the cable to fan tray 6.
If you disconnected the cable to fan tray 6 as part of a hot-plug procedure for fan tray 5 and you are immediately replacing fan tray 5, do not reconnect the cable to fan tray 6 until you install fan tray 5.
The procedure "How to Install a Motherboard Fan Tray" in the Sun Fire 880 Server Service Manual is incorrect. On systems configured with the redundant cooling fans option, installing motherboard fan tray 5 (the primary motherboard fan tray) requires that you disconnect the cable for motherboard fan tray 6.
The procedure should include the following steps in this order:
1. Locate the slot into which you want to install the motherboard fan tray.
a. If you are installing fan tray 5 and fan tray 6 is installed, disconnect the cable to fan tray 6.
2. Align the fan tray to be installed with its plastic guide in the chassis.
3. Slide the fan tray into the chassis.
4. If you disconnected the cable to fan tray 6 to install fan tray 5, reconnect the cable to fan tray 6.
Continue the procedure at step 4 in the Sun Fire 880 Server Service Manual.
For more information, see "About Fan Trays" in the Sun Fire 880 Server Owner's Guide or the Sun Fire 880 Server Service Manual.
The procedure "How to Install a Disk Drive" as documented in the Sun Fire 880 Server Service Manual and the Sun Fire 880 Server Owner's Guide is incomplete. The following task should be included in the procedure after Step 10:
11. If you are installing more than one disk drive as part of a hot-plug procedure, wait for the green light on the drive you just installed to light steadily (not flashing) before installing another drive.
The procedure "How to Install the FC-AL Disk Cage" as documented in the August 2001 Sun Fire 880 Server Service Manual is incomplete. The following task should be included as Step 9c:
9c. If a Sun StorEdge PCI Dual Fibre Channel Host Adapter card is installed to control Loop B of the FC-AL disk backplane, connect the card's FC-AL data cable to the base backplane at connectors C(J01100) and D(J01101).
DIMMs for the Sun Fire 880 are no longer available in kits of four DIMMs. The following table lists the single DIMMs and their part numbers for the Sun Fire 880.
Appendix A of the August 2001 Sun Fire 880 Server Service Manual lists incorrect part numbers for the following field-replaceable units (FRUs):
The correct FRU part numbers are as follows.
Note - Disk drive capacities lower than 36 Gbytes are not supported on the Sun Fire 880 server. |
The table for CPU Side Components transposes the part numbers for items 2 though 4. The corrected part numbers are as follows.
The cabling instructions for the Loop B PCI FC-AL data cable are incorrect due to a late change to the labeling scheme on the cable's connectors.
The correct cabling information is provided below.
Note that the following sections of the August 2001 Sun Fire 880 Server Service Manual are affected by this change:
The procedure "How to Install the Expansion FC-AL Backplane" as documented in the August 2001 Sun Fire 880 Server Service Manual is incomplete. The following should be included in the procedure before the reconfiguration boot instructions in the "What Next" section.
1. After installing an expansion backplane, power on the system and bring the system up to the ok prompt.
2. Allow the system to remain at the ok prompt for at least 10 minutes to ensure that the two backplanes are loaded with the same version of firmware.
The system automatically synchronizes the firmware versions between the two backplanes.
3. After the required waiting period, boot the system to single-user mode.
ok boot -s |
4. To verify that the firmware synchronization process has successfully completed, type the following luxadm subcommand:
# luxadm display enclosure_name |
Where enclosure_name is the enclosure name assigned to the Sun Fire 880 internal storage array--by default, FCloop. If you need to verify the enclosure name first, use the luxadm probe subcommand.
The output of the display subcommand shows the status of each SSC100 in the system. The following is an excerpt of sample output for a dual-backplane system.
SSC100's - 0=Base Bkpln, 1=Base LoopB, 2=Exp Bkpln, 3=Exp LoopB SSC100 #0: O.K.(9228/ 3A20) SSC100 #1: O.K.(9228/ 3A20) SSC100 #2: O.K.(9228/ 3A20) SSC100 #3: O.K.(9228/ 3A20) |
Verify that each SSC100 processor displays an "O.K." status and that each displays the same firmware version in parentheses. If so, the firmware synchronization process has successfully completed. Otherwise, wait another two minutes or so and repeat this step.
Note - For more information about the luxadm utility, see Platform Notes: Using luxadm Software, part of the Solaris on Sun Hardware AnswerBook2 Set on the Supplement CD. |
5. Once the firmware synchronization process is complete, you can restore the system to multi-user mode.
# init 3 |
The command for displaying POST results in the August 2001 Sun Fire 880 Server Service Manual is incorrect. To display the results of POST testing, type show-post-results at the ok prompt. The following is an abbreviated example of the command and output.
ok show-post-results
CPU0/Memory: OK
CPU1/Memory: OK
...
The obdiag-trigger and post-trigger information in "How to Enable ASR" in the Sun Fire 880 Server Owner's Guide is misleading. Step 2 recommends setting the obdiag-trigger variable to power-reset, error-reset, or soft-reset. However, setting the variable to power-reset requires the user to power cycle the system in order for the system to deconfigure the failed hardware component.
Since the purpose of ASR is to provide for an automatic system recovery from certain types of hardware failures, setting obdiag-trigger variable to power-reset does not allow for an automatic recovery from a hardware failure. To enable ASR for an automatic recovery from a hardware failure, set the obdiag-trigger variable to error-reset or soft-reset as shown in the following example.
ok setenv obdiag-trigger all-resets |
The default testing level of OpenBoot diagnostics tests is min. At this level of testing, OpenBoot diagnostics tests might not detect a failed hardware component. If you have enabled ASR on the system but have not set the OpenBoot diagnostic test level to max, the system might not recognize and disable a failed hardware component. In this case, during an ASR intiated reboot the system might not detect the failed component. The system will ASR again as soon as it reboots because the failed component has not been diasabled by ASR. In this case, the system enters a cycle of failed reboots followed by ASR resets.
To set the diag-level variable to max,type the following at the ok prompt.
ok setenv diag-level max |
Note - The ASR feature is not activated until you enable it at the system ok prompt. For more information about ASR, see the Sun Fire 880 Server Owner's Guide. |
To fully enabled ASR in the system, configure the OpenBoot settings as follows.
The procedure for hot-plugging a PCI card on a Sun Fire 880 Server in the Sun Fire 880 Dynamic Reconfiguration User's Guide (816-1465-10) does not fully explain the syntax for the ifconfig command. To prepare a PCI card for removal using the ifconfig command, you must first use the down option and then the unplumb option. To prepare a PCI card for installation using the ifconfig command, you must first use the plumb option and then the up option. The ifconfig syntax for removing a PCI card is as follows.
# ifconfig hme2 down
# ifconfig hme2 unplumb
The ifconfig syntax for installing a PCI card is as follows.
# ifconfig hme2 plumb
# ifconfig hme2 up
See the Sun Fire 880 Dynamic Reconfiguration (RSC) User's Guide for more information on Sun Fire 880 hot-plug operations.
Note - Not all PCI cards support PCI hot-plug operations. See the Sun Fire 880 Server Owner's Guide and Sun PCI Cards That Support PCI Hot-Plug of these Product Notes for more information on hot-plugging PCI cards. |
On a Sun Fire 880 system, the attachment points for PCI cards are not displayed as they are described the Sun Fire 880 Dynamic Reconfiguration User's Guide (816-1465-10). See PCI attachment points should be generic (BugID 4388625) in these Product Notes for a complete description of the current attachment point ids.
This section describes bugs and anomalies associated with the Sun Fire 880 server. In many cases, software patches that provide fixes for these bugs are available. Visit the SunSolve Online web site, or contact your Sun authorized service provider for information about patch availability. For more information, see Required And Recommended Software Patches.
The description of the attachment point ids (Ap_Id) for PCI cards in the Sun Fire 880 Dynamic Reconfiguration User's Guide, shown in the following example, is incorrect for Sun Fire 880 systems running the Solaris 8 operating environment.
# cfgadm
Ap_Id Type Receptacle Occupant Condition
pci0 pci-pci/hp connected configured ok
pci1 pci-pci/hp connected configured ok
pci2 pci-pci/hp connected configured ok
pci3 unknown connected configured ok
pci4 display/hp connected configured ok
pci5 pci-pci/hp connected configured ok
pci6 mult/hp connected configured ok
pci7 unknown connected configured ok
pci8 ethernet/hp connected configured ok
Currently, the attachment points for PCI cards on Sun Fire 880 systems running the Solaris 8 07/01 or 8 10/01 operating system are as follows.
# cfgadm
Ap_Id Type Receptacle Occupant Condition
pcisch0:hpc1_slot2 pci-pci/hp connected configured ok
pcisch0:hpc1_slot3 pci-pci/hp connected configured ok
pcisch0:hpc1_slot4 pci-pci/hp connected configured ok
pcisch0:hpc1_slot5 unknown connected configured ok
pcisch2:hpc2_slot2 display/hp connected configured ok
pcisch2:hpc2_slot3 pci-pci/hp connected configured ok
pcisch2:hpc2_slot4 mult/hp connected configured ok
pcisch3:hpc0_slot1 unknown connected configured ok
pcisch3:hpc0_slot2 ethernet/hp connected configured ok
When using cfgadm commands, use the attachment point id displayed by the cfgadm command. The following example shows the current attachment point ids and correct cfgadm command syntax for a Sun Fire 880 system.
# cfgadm
Ap_Id Type Receptacle Occupant Condition
pcisch0:hpc1_slot2 pci-pci/hp connected configured ok
...
pcisch3:hpc0_slot2 ethernet/hp connected configured ok
# cfgadm -c disconnect pcischO:hpc1_slot2
On a Sun Fire 880 server, attempting to stop, picld, a system environmental monitoring process, results in a segmentation fault followed by picld restarting in degraded mode.
Note - For further information about this issue, see the SunSolve Online web site. |
Copyright © 2003, Sun Microsystems, Inc. All rights reserved.