| Solaris 8 2/04 Release Notes Supplement for Sun Hardware |    
|
| Solaris 8 2/04 Release Notes Supplement for Sun Hardware |
|
| C H A P T E R 3 |
|
Open Issues |
This chapter discusses:
If your system is configured with Sun StorEdge SAN 4.x software and firmware, determine whether the latest upgrades have been installed. More information is available at:
http://www.sun.com/storage/san/
If you have SunScreen
SKIP 1.1.1 software currently installed on your system, you should remove the SKIP packages before installing or upgrading to the
Solaris 8 2/04 operating environment. The packages you should remove are: SICGbdcdr, SICGc3des, SICGcdes, SICGcrc2, SICGcrc4, SICGcsafe, SICGes, SICGkdsup, SICGkeymg, SICGkisup.
The following bugs have been filed against PCMCIA device support in this Solaris release.
If there is a PCMCIA I/O card in a slot that has been accessed and the card driver is unloaded (as occurs during modunload -i 0 at the end of multi-user boot) the system appears to hang.
Workaround: Removing the card may bring the system back.
When booting the Solaris 8 operating environment initially with a 64-bit kernel, the /dev/term/pc0 hardware device node for PC card serial device may not be created.
Workaround: Boot the system with the 32-bit kernel first, then reboot with the 64-bit kernel.
This release of the Lights Out Management (LOM) software is currently supported on Sun Fire V100, Sun Fire V120, Netra T1, Netra X1, Netra t1, Netra t1400/1405 servers. For configuration and operation instructions for Lights Out Management software, see the relevant chapter in the User Manual that came with the supported server platforms.
1. Insert the Supplemental CD and change to the LOMlite 2.0 software directory:
2. Add the LOMlite 2.0 packages, responding y to all questions:
3. Verify that the LOMlite 2.0 packages were installed:
The LOMlite2 software is now installed. Continue by adding LOMlite2 Patch 110208 as described in the next section.
Ensure you have the latest revision of Patch 110208 from SunSolve (revision -13 or later).
|
Note - The LOMlite2 software does not function unless this patch is added. |
1. Confirm the base LOMlite2 firmware version.
Type #. to change to the lom> prompt, then run version.
#. lom>version LOM version: v4.4-LW2+ LOM checksum: 387e LOM firmware part# 258-7939-12 Microcontroller: H8S/2148 LOM firmware build Aug 30 2001 18:02:40 Configuration rev. v1.2 lom>console |
The LOM firmware version must not be earlier than 4.5.
2. Type console at the lom> prompt to return to the # prompt.
3. Confirm that the device node is present (via PROM 4.2.4):
4. Copy the patch to a temporary directory (/var/tmp).
Confirm that the patch is present:
7. Confirm that the patch has been loaded:
8. To upgrade the firmware, type:
When prompted, type C to continue.
|
Note - This process takes several minutes. Do not turn off the system while the update is progressing. |
9. Press Return to redisplay the UNIX prompt.
|
Note - If the system displays the ok prompt, type go to return to the UNIX prompt. |
10. Check the functionality by running the lom -a command:
The software is now installed and the system is ready for use.
A segmentation fault error occurs when invoking the Solstice DiskSuite 4.1.2 metatool on a Sun Enterprise 6000 system with Alternate Pathing 2.3 or 2.3.1 already installed and configured on a SPARCstorage Array 100 or SPARCstorage Array 200.
After invoking the metatool, the following message is displayed:
root@[/]>metatool & [1] 2569 root@[/]>Initializing metatool... Done. Discovering drives and slices... metatool: Segmentation Fault |
Workaround: Move /usr/lib/libssd.so.1 and /usr/lib/libap_dmd.so.1 to /usr/sadm/lib/lvm.
When the ToolTalk Database Server (/usr/openwin/bin/rpc.ttdbserverd) is enabled, remote and local attacks are possible. These attacks can potentially give root access to the system. Attacked systems can be identified by the presence of garbage files in the root directory that begin with an A.
Workaround: Remove the rpc.ttdbserverd line from the inetd.conf file. Disable the service by using the chmod a-x command on
/usr/openwin/bin/rpc.ttdbserverd. Check with your Sun representative about future patches.
VERITAS File System (tm) 3.5 installation fails on systems running Solaris 8 2/04 because it depends on patch 108901, which was obsoleted by patch 108528-24. The VRTSvxfs installation package must be modified to remove the dependency of patch 108901-03 or later. The following error message appears:
Workaround: Retrieve the modified files from the Veritas website:
http://seer.support.veritas.com/docs/264856.html
Follow the instructions listed with the files.
When attempting to install the third-party Veritas 3.4 filesystem vxvm using the pkgadd utility, the VRTSvxfs package might fail installation with an error message that reads, in part:
Veritas has fixed this bug in the installation script. A new revision of the Version 3.4 vxfs is available from Veritas which fixes this problem. The Version 3.5 vxfs product has also corrected this problem. Both of these versions are available directly from the Veritas web site:
The following issues apply to the SunVTS 5.1 PS5 product.
The SunVTS 5.1 PS5 software is compatible with the Solaris 8 2/02, Solaris 8 HW 12/02, Solaris 8 HW 5/03, Solaris 8 HW 7/03, Solaris 8 2/04, Solaris 9, Solaris 9 9/02, Solaris 9 12/02, Solaris 9 4/03, Solaris 9 8/03, Solaris 9 12/03, and Solaris 9 4/04 operating environments. The following new features and tests are added to the SunVTS 5.1 PS5 release:
|
Note - The S24 Frame Buffer Test (tcxtest) might be discontinued in a future release of SunVTS. |
|
Note - The SunPCi |
Refer to the SunVTS 5.1 Patch Set 5 Documentation Supplement for details about the new features and tests offered in this release.
|
Note - The SPARCstorage® Array Controller Test (plntest) is discontinued in Solaris 9 8/03. |
The online help documentation available with the SunVTS 5.1 software includes a chapter describing the RAM test (ramtest). This test is supported only in SunVTS 5.1 Patch Set 4 (PS4) onward.
This section lists corrections to errors in the Supported Test Modes tables of the following test chapters in the SunVTS 5.1 Test Reference Manual:
Compact Disk Test (cdtest)
CPU Test (cputest)
Disk and Floppy Drives Test (disktest)
DVD Test (dvdtest)
ECP 1284 Parallel Port Printer Test (ecpptest)
Sun StorEdge A5x00 Test (enatest)
Functional and Online test modes are supported in SunVTS 5.1.
Sun StorEdge 1000 Enclosure Test (enctest)
Environmental Test (env2test)
Online and Exclusive test modes are not supported in SunVTS 5.1.
Environmental Test (env4test)
Floating Point Unit Test (fputest)
Online and Exclusive test modes are supported in SunVTS 5.1.
Cache Consistency Test (mpconstest)
Multiprocessor Test (mptest)
Qlogic 2202 Board Test (qlctest)
Only Exclusive and Connection test modes are supported in SunVTS 5.1.
Serial Ports Test (sptest)
System Service Processor Test (ssptest)
SunHSI Board Test (sunlink)
System Test (systest)
Tape Drive Test (tapetest)
S24 Frame Buffer Test (tcxtest)
In Appendix A, SunVTS Window and Dialog Box Reference, Table A-11 indicates the incorrect default value of 0 (zero) for the Max Errors field of the example systest Test Parameter Options dialog box in Figure A-11. The correct default value of the Max Errors field of the systest Test Parameter Options dialog box is 1 (one) as indicated in Figure A-11.
Also in Appendix A, Table A-13 indicates the incorrect default value of 0 (zero) for the Max System Errors field of the example systest Threshold Options dialog box in Figure A-13. The correct default value of the Max System Errors field of the systest Threshold Options dialog box is 1 (one) as indicated in Figure A-13.
You might encounter an installation problem when you attempt to install SunVTS with an installation program other than the pkgadd command as described in the following subsections.
If SunVTS 5.1 software is installed in the default /opt directory, you cannot install a subsequent SunVTS 5.1 Patch Set release in a different directory on the same system. When this duplicate installation is attempted with pkgadd, the following error message occurs:
The reason for this error is that the base package revision is the same for both SunVTS 5.1 and any subsequent SunVTS 5.1 Patch Set release. When a SunVTS 5.1 Patch Set release is installed in the default /opt directory which already has SunVTS 5.1 software installed, the installation completes successfully with the following warning message:
This appears to be an attempt to install the same architecture and version of a package which is already installed. This installation will attempt to overwrite this package. |
From SunVTS 5.1, the SunVTS software depends on XML packages that are not included with the Solaris 8 2/02 distribution. For Solaris 9, all of the prerequisite packages are installed with the End User Solaris Software Group as a minimum.
You may install SunVTS 5.1 on Solaris 8 2/02, Solaris 8 HW 12/02, Solaris 8 HW
5/03, Solaris 8 HW 7/03, or Solaris 8 2/04. If you choose to install SunVTS 5.1 or later on any of these releases, you must separately install the required XML packages (SUNWlxml[32-bit] and SUNWlxmlx[64-bit]) from either the Solaris 8 HW 12/02, Solaris 8 HW 5/03, Solaris 8 HW 7/03, or Solaris 8 2/04 Software Supplement CD-ROM.
Install the minimum End User Solaris Software Group.
Install the XML packages SUNWlxml(32 bit) and SUNWlxmlx(64 bit) from either the Solaris 8 HW 12/02, Solaris 8 HW 5/03, Solaris 8 HW 7/03, or Solaris 8 2/04 Software Supplement CD-ROM.
|
Note - These XML packages are not available in the Solaris 8 2/02 distribution. |
Install SUNWzlib(32 bit) and SUNWzlibx(64 bit) packages from the Entire Solaris Software Group from either the Solaris 8 2/02, Solaris 8 HW 12/02, Solaris 8 HW
5/03, Solaris 8 HW 7/03, or Solaris 8 2/04 Software CD-ROM.
Install the optional SUNWcpc(x)/SUNWcpcu(x) packages only if you want to enable displaying certain performance counters for certain CPU and memory tests. Otherwise, these packages need not be installed.
The SUNWcpc(x)/SUNWcpcu(x) packages are included in the Entire Solaris Software Group for Solaris 8 2/02, Solaris 8 HW 12/02, Solaris 8 HW 5/03, Solaris 8 HW 7/03, and Solaris 8 2/04. These packages are not included in the End User Solaris Software Group for Solaris 8 2/02, Solaris 8 HW 12/02, Solaris 8 HW 5/03, Solaris 8 HW 7/03, nor Solaris 8 2/04.
To install SunVTS for Solaris 9 and later releases:
Install the End User Solaris Software Group (or any software group).
The XML packages (SUNWlxml[32-bit] and SUNWlxmlx[64-bit]) and the SUNWzlib packages (SUNWzlib[32 bit] and SUNWzlibx[64 bit]) are part of the End User Solaris Software Group (and Entire Solaris Software Groups) for Solaris 9 releases.
Install the optional SUNWcpc(x)/SUNWcpcu(x) packages only if you want to enable displaying certain performance counters for certain CPU and memory tests. Otherwise, these packages need not be installed.
The SUNWcpc(x)/SUNWcpcu(x) packages are part of the Entire Solaris Software Group and not the End User Solaris Software Group.
Web Start 2.0 may not install SunVTS on systems that do not have the Solaris 64-bit environment installed. Web Start 2.0 removes the SunVTS 32-bit packages when the SunVTS 64-bit packages cause the installation to suspend.
Workaround: Use the pkgadd command to install the 32-bit SunVTS packages as described in the SunVTS 5.1 User's Guide.
When you install SunVTS using Web Start 2.0, you are not prompted to enable the Sun Enterprise Authentication Mechanism (SEAM) Kerberos v5, SunVTS security feature. The installation defaults in a way that installs SunVTS without this high level of security. If you do not want the high-level security, there is no problem.
Workaround: To enable the high-level SEAM security, use the pkgadd command to install SunVTS packages as described in the SunVTS 5.1 User's Guide.
When you attempt to install SunVTS using Web Start 2.0, you are unable to change the directory where SunVTS is installed. SunVTS will be installed in /opt.
Workaround: Use the pkgadd -a none command to install SunVTS in the directory of your choice as described in the SunVTS 5.1 User's Guide.
Use the same tool or utility for installation and removal of the SunVTS software. If you use pkgadd for installation, use pkgrm to uninstall; if you use Web Start for installation, use the Product Registry to uninstall.
When performing the following SunVTS tests on Sun Fire E25K systems with 72 UltraSPARC IV processors (144 CPU IDs), the following issues might occur:
systest - A Bus Error (core dumped) error might occur (Bug ID 4981458)
mptest - The test could hang (Bug ID 4982924)
mpconstest - The test could fail because of too many CPU IDs (Bug ID 4982944)
cmttest - The test times out (Bug ID 4982948)
cmttest - The test fails (Bug ID 4981014)
Workaround: Look for the latest version of Patch ID 116042 at:
cmttest is not currently supported for Netra servers with Jaguar UltraSPARC IV CMTs (Chip Multiprocessors).
When performing systest on Sun Fire 15K systems with 72 UltraSPARC IV processors (144 cpuIDs), a Bus Error (core dumped) error might occur.
On Sun Fire 15K systems, adding new boards might cause some of the processor and memory related tests to perform ineffectively. Specifically, cmttest might fail to recognize the CMT processors on the new board. Similar failures might also occur in l2sramtest, l1dcachetest, dtlbtest, ramtest, bustest, mptest, and fputest.
Workaround: Reboot the system after adding a new board.
pfbtest might fail when performed in the Gnome desktop environment on a Sun XVR-100 graphics accelerator if the test is performed in the default console window.This failure does not occur in the Solaris 8 2/02 and Solaris 8 HW 3/03 operating environments. If this failure occurs, you see an error message similar to the following:
An ecpp/parallel port driver issue might cause ecpptest to fail. The failure rate is three out of 130 machines; the time to fail is about 10 hours.
Sun Blade 100 and 150 systems with SunVTS can produce data corruption errors when the serial port controller (southbridge) is also handling other heavy traffic such as data access from and to the IDE hard disk.
With SunVTS simultaneously running sutest and disktest on Sun Blade 100 and 150 systems, you might see sutest report failures similar to the following:
The difference between the expected (exp) and the observed (obs) values is always 0x10.
The long term technical fix for this problem will involve the installation of a patch or driver to be released in the near future. The root cause of this failure is still under investigation.
Workaround: Do not perform sutest and disktest simultaneously.
If you perform the Sun XVR-4000 Graphics Accelerator Test (zulutest) on a system that was powered on without running X-Windows, you must bring up X-Windows on the Sun XVR-4000 Graphics Accelerator device under test and kill the X-Windows process before performing zulutest. Otherwise, the Convolve subtest might fail, and other subtests might also fail.
|
Note - You must enable multisampling with the fbconfig command before performing the following workaround. To perform zulutest with X-Windows (CDE) the following workaround is not necessary. |
Workaround: To bring up X-Windows on the Sun XVR-4000 Graphics Accelerator device under test, enter the following command:
It takes 30 to 45 seconds before Xsun comes up. To kill the Xsun process, enter the following command:
Once the Xsun process is killed, the zulutest can be performed without the incorrect subtest errors.
The Sun XVR-4000 Graphics Accelerator cannot perform video read back in Interlaced and Stereo modes because the Convolve subtest cannot keep up.
For zulutest to be able to perform the Convolve subtest, multisampling must be enabled.
The command pkginfo -c sunvts does not produce any output in SunVTS 5.1. This situation correctly implements the -c option of the pkginfo command.
Workaround: Use the following command to receive SunVTS 5.1 package information
You can also use either of the following commands to receive additional SunVTS 5.1 package information:
Performing sutest on a port that is being used as console causes sutest to fail.
Workaround: Do not perform sutest on a port that is being used as a console.
Using afbtest in a Gnome environment might stop the Xserver when the screen-saver is set to disable. If the Xserver is stopped, afbtest might fail immediately with the following error messages:
Workaround: Type the xhost + command on the machine being tested and continue the test; the afbtest will pass.
SunVTS may fail in multi-user mode when semaphores are created improperly.
Workaround: Kill and restart picld after the machine reaches multiuser mode, but before running SunVTS:
An error might occur when jnifctest is performed on a system with 5 JNI ports. This error occurs on one card and the error can be seen using the data pattern 0x00000000. The value returned for the fcio_errno statistic is 65 (decimal).
Workaround: Turn off the self-test or perform the test on one port only.
When performing disktest, the test might probe and premount the Solstice DiskSuite (SDS) partitions as mirrors.Workaround: Set the BYPASS_FS_PROBE parameter to 0.
If the /var/opt/SUNWvts/sched_manage directory is not present, schedules cannot be created.
Workaround: Create the /var/opt/SUNWvts/sched_manage directory before creating schedules.
The qlctest 10-bit/1-bit internal loopback subtests might fail when performed on Sun Fire V880 platforms. This problem is fixed in the latest SAN Foundation software.
Workaround: Install the Sun StorEdge SAN Foundation Software 4.2 or later from: http://www.sun.com/storage/san/
This document deals with Sun Remote System Control (RSC) 2.2 hardware and software issues. For complete information about using RSC, see the Sun Remote System Control (RSC) 2.2 User's Guide.
Several new features are available in RSC 2.2:
The RSC GUI requires an updated version of the Java Runtime Environment, Java 2 Standard Edition (J2SE) Runtime Environment Version 1.3.0_02 or greater. You can download the appropriate version from one of these Web sites:
Client support has been added for the Microsoft Windows 2000 operating environment. RSC 2.2 does not support Windows 95.
Sun Fire V480 servers include a new hardware feature, a Locator LED on the system's front and rear panels. RSC software allows you to toggle the state of these LEDs to help identify a particular system that may be located in a rack with other servers.
Support for a maximum of 16 RSC user accounts has been added. A maximum of 10 users can be logged in at one time. However, the increased number of user accounts does not affect the limitation of five concurrent telnet or GUI login sessions per server.
RSC software is included as part of the default installation set for this Solaris Supplement CD. You should install RSC server components on a compatible Solaris server only; you can install the client software on any computer that meets the Solaris or Windows operating environment requirement. You must install and configure the RSC software before you can use RSC.
Important: Before upgrading from a previous version of RSC server software or reinstalling the software, log in to the server as superuser and back up your configuration data using the following commands:
Use a meaningful file name that includes the name of the server that RSC controls. After installation, you can refer to this file to restore your configuration settings if necessary. Reverting to a previous version of RSC server software after installing version 2.2 is not recommended. However, if you do revert, you will need to restore your configuration information and also power cycle the server.
You can install the RSC 2.2 server software package, SUNWrsc, on:
A Sun Fire V880 server running the Solaris 8 7/01 operating environment or another Solaris version that supports the RSC 2.2 product
A Sun Fire 280R server running the Solaris 8 1/01 operating environment or another Solaris version that supports the RSC 2.2 product
You can install the RSC 2.2 client software packages on:
Any other computer running the Solaris 2.6, Solaris 7, or Solaris 8 operating environment. The packages are SUNWrscj (GUI) and SUNWrscd (documentation).
Client computers require Java 2 Standard Edition (J2SE) Runtime Environment Version 1.3.0_02 or a subsequent 1.3.x version to run RSC 2.2 software. RSC 2.2 software does not run using J2SE Runtime Environment Version 1.2.x. You can download the appropriate version from one of these Web sites:
Installation on the Solaris operating environment places the Sun Remote System Control (RSC) 2.2 User's Guide in the location /opt/rsc/doc/locale/pdf/user_guide.pdf. Installation on the Windows operating environment places the User's Guide in the location C:\Program Files\Sun Microsystems\Remote System Control\doc\locale\pdf\user_guide.pdf.
The following sections describe Sun Remote System Control (RSC) 2.2 issues.
This section describes issues that affect RSC running on all platforms (Sun Enterprise 250, Sun Fire 280R, Sun Fire V880, and Sun Fire V480 servers):
If the RSC variables page_enabled and mail_enabled are set to true and multiple alert messages are generated within a short interval, the first message is delivered in a timely fashion but each subsequent message issued during the interval is delayed by 3-4 minutes.
When the RSC card is running on battery power, the keyswitch slot in the RSC GUI displays as a gray dot, and the mouse-over text on the keyswitch reads Current Keyswitch Position Unknown.
When configuring the page_info1 or page_info2 fields, you may use any digit or the alphanumeric characters #, @, and , (comma) when specifying a pager phone number, but the PIN area may only contain digits (0-9). In the RSC GUI, the online help for this function is incorrect. For more information about how to configure RSC to work with a pager, refer to the Sun Remote System Control (RSC) 2.2 User's Guide.
If diag-switch? is set to true and you use the bootmode -u command to reboot your workstation, rsc-console will revert to the serial (tip) connection after Solaris restarts, even if you have previously redirected the console to RSC.
If this occurs, manually redirect the console ouput to RSC again after the reboot operation has completed. Refer to the Sun Remote System Control (RSC) 2.2 User's Guide for more information.
If the system's state changes from battery power to standby power and the keyswitch is in the off position, the RSC software should generate a warning that you cannot turn the system power back on when you issue a poweron command from either the CLI or the GUI. The RSC software should issue this warning. You need to move the keyswitch back into the on position.
If this happens, use the resetrsc command.
This intermittent problem has been observed on Sun Fire V880 servers running OpenBoot PROM version 4.4.6. Occasionally, the bootmode -u command fails to redirect the console to RSC. If this happens, use the resetrsc command.
On servers running OpenBoot PROM software version 4.4.3, an RSC-initiated system reset produces the following message:
ERROR: RSC-initiated Reset
This message serves as a warning-level message only; no action is required.
If you install the RSC client on the Japanese version of Microsoft Windows 98, RSC does not start again once you exit from the client and the javaw stack error appears, unless you reboot the system. This situation occurs only on the initial version of Windows 98, and does not happen on other versions of Microsoft Windows (95, 98 Second Edition, NT).
Workaround: Download and install the Microsoft IME98 Service Release 1 (IME98-SR1) from the Microsoft Website. The crash does not occur after IME98-SR1 has been installed.
This section describes issues that affect RSC running on Sun Fire 280R, Sun Fire V880, and Sun Fire V480 servers.
CAUTION: Removing or installing the RSC card while the system has the AC power cord connected could damage your system or your RSC card. Only qualified service personnel should remove or replace the RSC card. Contact your qualified service representative to perform this service operation.
Before you follow the procedures in the Sun Fire 280R Server Service Manual or Sun Fire 880 Server Service Manual to remove or install the RSC card, perform this procedure to ensure that there is no AC power present in the system.
1. Shut down and halt the system.
2. With the system at the ok prompt, turn the keyswitch to the Off position.
Standby power is still present in the system at this point.
3. Disconnect all AC power cords from their back panel receptacles.
This ensures that there is no standby power voltage present in the system.
4. Follow the procedure you require in your service manual.
RSC generates the following alerts on a Sun Fire 280R or Sun Fire V880 server when the RSC card begins battery use after a power interruption:
00060012: "RSC operating on battery power."
RSC generates the following alerts when the host system has shut down from RSC. The messages appear in the log history.
00040000: "RSC Request to power off host."
00040029: "Host system has shut down."
If you shut down the system using the keyswitch, or by using the OpenBoot PROM poweroff command, the above alert 00040029 is the only alert displayed.
These alerts are not documented in the Sun Remote System Control (RSC) 2.2 User's Guide.
When the RSC hardware changes state from battery power to standby power, the Locator LED on the Sun Fire V480 appears illuminated in the GUI only. It does not illuminate on the system.
If this situation happens, use the resetrsc command.
This section describes issues that affect RSC running on Sun Fire 280R servers only. See the Sun Fire 280R Server Product Notes for other Sun Fire 280R server issues.
The xir command does not bring the server to the ok prompt as expected. This issue may have been resolved for your server type; see your hardware platform release notes.
In rare instances, the system may bypass the RSC card during startup. To check whether the system booted and is online, use the ping command to see if the card is alive, or log in using telnet or rlogin. If the system is not connected to the network, establish a tip connection to the system. (Be sure that console I/O is not directed to the RSC card.) Use the tip connection to view boot messages on the troubled system, or reboot the system. For help in diagnosing the problem, see your hardware Owner's Guide.
When you power on the system, it may report a false internal drive fault that is recorded in the Sun Remote System Control (RSC) log history.
If the error is reported by RSC, you should disregard the report if the system boots successfully to the Solaris operating environment. In most cases the erroneous fault does not reappear. You can verify the disk after the boot process by using the fsck utility.
|
Note - Any disk drive error message reported by the Solaris operating environment is a real disk drive error. |
If a disk fault is reported at the ok prompt and the system fails to boot to the Solaris operating environment, there may be a problem with the disk drive. Test the disk drive with the OpenBoot Diagnostics tests documented in the "Diagnostics, Monitoring, and Troubleshooting" chapter in the Sun Fire 280R Server Service Manual.
After a cold restart or after powering on the system, the RSC command rscadm resetrsc fails; this is a known condition. You need to reset the host system for the command to function correctly.
There are three ways you can reset the host. Use one of the following commands:
The RSC rscadm resetrsc command will now function correctly.
This section describes an issue that affects RSC running on Sun Fire V880 servers only.
The xir command does not bring the server to the ok prompt as expected. This issue may have been resolved for your server type; see your hardware platform release notes.
This section describes issues that affect RSC running on Sun Enterprise 250 servers only. See the Sun Enterprise 250 Server Product Notes for other Sun Enterprise 250 server issues.
Support for a maximum of 16 RSC user accounts has been added for RSC 2.2. However, Sun Enterprise 250 servers continue to be limited to four RSC user accounts because of hardware limitations.
Do not issue the fsck command from the redirected RSC console.
Reset the system's input-device and output-device settings to ttya. Then reboot the system and access the system through its local console or terminal and execute the OpenBoot PROM fsck command directly.
The command boot -s does not work from the RSC console.
Reset the system's input-device and output-device settings to ttya. Then reboot the system and access the system through its local console or terminal and execute the boot -s command directly.
In order for changes to the RSC configuration variable serial_hw_handshake to take effect, the server must be rebooted. This also affects the Enable Hardware Handshaking check box in the RSC graphical user interface. This limitation is not stated in the documentation.
In the Sun Enterprise 250, the power supplies are numbered 0 and 1, but the RSC GUI refers to them as Power Supply 1 and Power Supply 2 in the event log and in alerts.
When calling luxadm remove_device /dev/rdsk/c1t1d0s2, the system asks for confirmation and reports that the disk is spinning down and being taken offline. When the disk is removed, pressing Return causes the system to report that the device links are removed.
However, the device links are not removed. The disk can then be re-inserted and used without running the luxadm insert_device command. When the command is subsequently run, the system reports that no new devices were added and the following message is produced at the console:
To work around, enter the following when luxadm remove_device claims to have completed successfully:
If four USB Iomega zip drives are attached, prtconf indicates that no driver is loaded after the system has booted. drvconf starts a system probe that finds only two of the four USB zip drives.
Running cat on the /dev/usb/hub0 file causes the system to add the two zip drives.
Workaround: Do not use more than two zip drives in the current release.
The LOMlite2 reset-x command does not currently reset the CPUs. Look for this functionality in a later revision of the OpenBoot PROM.
A warning message may appear when booting a Netra ct 800 or Sun Blade 100 system which reads:
This message is benign and can be ignored.
The following documentation errors have been found in the Solaris 8 2/04 documents.
References to System Management Services (SMS) in the Solaris 8 2/04 Sun Hardware Platform Guide mistakenly list version 1.4. Instead, SMS 1.3 is supported in this release.
Non-English versions of this document might mention the PC Launcher software as included on the Supplement CD. However, this software is not supported in this Solaris release.
The Sun Enterprise 10000 SSP 3.5 User Guide, Sun Enterprise 10000 Dynamic Reconfiguration User Guide, and the Sun Enterprise 10000 DR Configuration Guide contain references to a document that is not available, namely the MPxIO Installation and Configuration Guide.
The MPxIO product is now called the Sun StorEdge Traffic Manager. Information on installing and configuring the Sun StorEdge Traffic Manager is provided in the Sun StorEdge Traffic Manager Software Installation and Configuration Guide, available on the Sun Network Storage Solutions website:
http://www.sun.com/storage/san
1. Go to http://www.sun.com/storage/san and click the "Sun StorEdge SAN3.0 release Software/Firmware Upgrades and Documentation" link.
3. Click the "Download STMS Install/Config Guide, Solaris/SPARC (.35 MB)" link.
Page 8: In the description of the snmpd memory leaks (Bug ID 4486454), you must be superuser on the main SSP to stop and start SSP daemons, not user ssp as indicated in the text.
Pages 75 and 99: The references to the Solaris 8 Installation Supplement are incorrect. Refer to the Solaris 8 Advanced Installation Guide for details.
The eri(7d) man page incorrectly states that the eri.conf file is in the
/kernel/drv/sparcv9/eri.conf file
The correct location is: /kernel/drv/eri.conf
The Computer Systems Installer CD automatically installs the Network Drivers from the Supplement CD. Drivers for the SunATM, SunHSI/S, SunHSI/P, SunFDDI/S, and SunFDDI/P cards are now installed by default when the Solaris 8 2/04 operating environment is installed. Error messages might be displayed if you do not have some of the corresponding hardware installed. You can ignore these error messages.
This spc driver does not support Dynamic Reconfiguration features in the Solaris 8 operating environment.
The default timeout value for the SCSI portion of the SunSwift PCI Ethernet/SCSI host adapter (X1032A) card does not meet the timeout requirements of the Sun SCSI DVDROM drive (X6168A). With marginal media, the DVD-ROM occasionally experiences timeout errors. The only exceptions are Sun Fire 6800, 4810, 4800, and 3800 systems, which overwrite the SCSI timeout value via OBP.
Workaround for other platforms: use the on-board SCSI interfaces or DVD-ROM compatible SCSI adapters, such as X1018A (SBus: F501-2739-xx) or X6540A (PCI: F375-0005-xx).
The PCI FDDI card has known hardware problems related to DTXSUSP (DMA TX process suspend interrupt) and DRXSUSP (DMA RX process suspend interrupt) bits in the MAIN_CONTROL and MAIN_STATUS registers. These hardware bugs minimally affect FDDI functionality under normal conditions. However, under a very heavy TX or RX load, the card posts an unexpected interrupt. The system posts the error message, "Interrupt level 6 is not serviced". After 20 such errors, the pcipsy driver shuts down interrupt processing for the FDDI and hangs FDDI.
The monitor may go blank if all of the following conditions occur:
Your frame buffer card is a Creator3D or an Elite3D.
The "Blank Screen" screen saver is used.
Stop-A is pressed while in the blank screen mode.
Workaround: If your screen saver is enabled, choose any pattern other than "Blank Screen."
At the command line login, systems with Expert3D or Expert3D-Lite cards may be unable to start a window system or dtlogin.
1. Remotely log into the troubled system as superuser.
2. Stop all ifbdaemon processes:
3. Start all ifbdaemon processes:
This software release features a dual USB framework. The original framework, released with Solaris 8 2/02, is provided for compatibility with all preexisting USB drivers. It supports USB 1.0 and USB 1.1 devices, and operates devices connected to on-board and other USB 1.0 or 1.1 ports.
The USBA 1.0 framework (new in Solaris 8 HW 5/03) is more evolved, and as such offers better performance and support for more devices than the original framework. It supports high speed USB 2.0 devices, and operates devices connected to the computer's USB 2.0 ports. Thus, the "dual framework" refers to the old USBA framework which supports only USB 1.0/1.1 devices, and the new USBA 1.0 framework which supports both USB 1.1 and USB 2.0 devices.
This dual framework release is supported on Sun Blade 1500 and Sun Blade 2500 systems. Systems pre-equipped with the IEEE 1394 and USB 2.x card support USB 2.0 through the card's USB 2.0 ports. Additionally, some aftermarket PCI USB 2.0 cards may work. For a listing of Solaris Verified PCI USB 2.0 cards for use with other platforms, see:
The USB console keyboard and mouse must be connected through an on-board USB port. This is required in order use them as console devices which can boot, halt and examine a system, and allow the window manager to start.
All other USB 1.1 devices can be connected to any USB port on the computer. Note, however, that any device connected to a USB 2.0 port on the computer can take advantage of the USBA 1.0 framework.
For best performance, connect USB 2.0 storage devices to a USB 2.0 computer port (optionally through a USB 2.0 hub). While USB 2.0 devices can function on other ports, they operate significantly slower when using USB 1.1.
USB 1.x devices do not function when connected through a USB 2.0 hub to a computer USB 2.0 port. You can connect the USB 2.0 hub to an on-board or other USB 1.1 port, use a USB 1.1 hub, or bypass the hub and plug the USB 1.x device directly into a USB port on the computer.
Please see http://www.sun.com/desktop/whitepapers.html for a link to a guide which discusses administration of USB devices on both frameworks. It is more complete than the System Administration Guide, which covers only the original framework.
This change was implemented to fix inconsistent behavior in previous releases, where some USB storage devices were seen as removable devices and others were not.
In general, USB devices now operate with a more consistent interface, and are easier to use:
The rmformat(1) program can now be used to format USB storage devices, instead of other, more complicated programs such as format(1M).
USB hard drives with a FAT file system can be mounted and used.
Non-root users can now access USB storage devices, since a root-privileged mount command is no longer needed. The volume manager mounts it and makes it available under /rmdisk. (See Caveats for restrictions.)
All USB storage devices are power managed, except for those which support log sense pages. Devices with log sense pages are usually SCSI devices connected through a USB to SCSI bridge device.
Error handling is more forgiving for removable media devices, with increased timeouts for recovery or preparing for the drive to become ready.
External removable media applications might not work correctly if they make assumptions about the size of the media or are not largefile aware. For example, a number field containing a disk size may overflow.
If a new USB device is connected while the system is down, a reconfiguration boot (boot -r) is required for the volume manager to find it for automatic mounting. If a new device is connected while the system is up, the volume manager must be restarted. Refer to vold(1M), scsa2usb(7D), the System Administration Guide, and Known Bugs for more information.
The format(1M) program must now be started in expert mode (-e) in order to see any USB storage device. In previous releases, those USB storage devices which were not seen as removable media could be seen without the -e switch. You must either answer questions about device geometry or furnish an entry for /etc/format.dat in order to utilize the program.
Some applications which make DKIOCEJECT ioctls may have returned ENOTTY for non-removable media. These now succeed with no effect because the device is treated as a removable media device.
Setting removable=false in /kernel/drv/scsa2usb.conf (original framework) or /kernel/drv/usba10_scsa2usb.conf (USBA 1.0 framework) for a particular device type overrides the forced treatment of that device type as removable media. Use this option if compatibility with previous releases is an issue. Please see these configuration files for more details.
For best performance, connect USB 2.0 storage devices to a USB 2.0 computer port (optionally through a USB 2.0 hub). While USB 2.0 devices can function on other ports, they operate significantly slower when using USB 1.1.
USB storage devices are not automatically seen by the vold process after a non-reconfigure reboot, due to a boot-time delay in starting the USB frameworks.
Workaround: use any of the four workarounds listed below.
As root, edit the /etc/init.d/initusb script. Change the entry sleep 60 to sleep 5. This alleviates the delay in starting the USB frameworks, and almost always causes the vold process to find the devices automatically; however, it also increases the boot time by a few seconds. If a device still is not found, hotplug it as in the second workaround listed.
Hotplug the USB storage device two minutes after the system has given the login prompt. This method must be repeated for each USB storage device after every reboot.
As root, stop and restart vold:
Do this after every reboot, two minutes after the system has given the login prompt.
As root, send a SIGHUP signal to the vold process to rescan for devices:
Do this after every reboot, two minutes after the system has given the login prompt.
|
Note - Due to a bug inthe vold process, CD-ROM drives might eject their media in response to this signal. |
The generic USB (ugen) driver only works with devices that support setting of configurations and alternate interfaces. Attempts to open data endpoints on devices which do not support these settings fail with an I/O error. Please see Chapter 9 of the USB 2.0 spec (http://www.usb.org) for more information.
As part of first-order identification (enumeration) to the operating system, USB devices say whether or not they provide a string descriptor (used to identify the device for messages). A system cannot use USB devices that say they provide such a descriptor and then fail to provide that descriptor.
Some USB devices fail to identify themselves within the system's allotted window. If this happens, the user sees the following error message:
Workaround: Hot-plugging the device again might successfully identify it.
USB floppy devices appear as removable media devices, as all other USB disks do. USB floppy devices are not managed by the fd (floppy) driver. Applications which issue ioctl(2) calls intended for the fd driver fail. Applications which issue only read(2) and write(2) calls succeed. Some other applications, such as SunPCI and rmformat, do work.
Volume Manager sees the USB floppy as a SCSI removable media device. Volume Manager makes the device available for access under /rmdisk.
File Manager does not fully support the USB floppy at this time. Floppies containing a ufs file system can be opened, renamed and formatted from the File Manager's Removable Media Manager. However, the only action that the Removable Media Manager can take with a floppy containing a pcfs file system is to open it. Floppies containing either type of file system can successfully drag and drop files.
Always perform a reconfigure boot (boot -r) when rebooting your system, if you plan to use a connected USB floppy device with Volume Manager and File Manager. When you hotplug a new USB floppy device, you must stop and restart vold in order for vold to find the newly connected device.
This section describes how to use a USB floppy without the File Manager.
After plugging in the device, find the device's name:
# cd /dev/rdsk # devfsadm -C # ls -l c*0 | grep usb lrwxrwxrwx 1 root root 55 Mar 5 10:35 c2t0d0s0 -> ../../devices/pci@1f,0/usb@c,3/storage@3/disk@0,0:a,raw |
Here, the device name shows as c2t0d0s0. The device can be accessed as raw device /dev/rdsk/c2t0d0s0, and as block device /dev/dsk/c2t0d0s0.
Floppies support both pcfs(7FS) and ufs file systems. Below are examples of how the different file systems are used. The examples assume a device of
/dev/[r]dsk/c2t0d0s0.
If multiple device names are displayed, determine which name corresponds to the USB floppy device:
Insert a floppy disk into the USB floppy device.
Cycle through each device, attempting to access each device (using od(1) on that device's raw partition, for example).
The USB floppy's indicator lamp lights up when you have found the floppy.
Floppies support both pcfs(7FS) and ufs file systems. The following commands are examples of how to use the two file systems. The examples assume a device of
/dev/[r]dsk/c2t0d0s0.
Use the following commands to create and use the floppy as a pcfs file system:
Use the following commands to create and use the floppy as a ufs file system:
The SunFDDI PCI board (FDDI/P) does not support diskless booting. The SunFDDI SBus board (FDDI/S) supports diskless booting on sun4m and sun4u platforms only.
Your system may experience heavy CPU usage or may hang if more than two Sun GigaSwift Ethernet cards are installed.
If you are running any version of the Solaris 8 software with an Expert3D or Expert3D-Lite card, you may experience problems after cpr runs.
Workaround: Install Patch 108576-15 or greater.
Performing a cfgadm -c disconnect (bus quiesce) command when the SCSI bus is unconfigured does not succeed.
Workaround: If the SCSI bus is unconfigured, use the cfgadm -c configure command before attempting the cfgadm -c disconnect command.
Using the cfgadm -c unconfigure command might fail if a qlc driver fails to detach. The following error message appears:
>cfgadm -c unconfigure pcisch6:sg8slot5 cfgadm: Component system is busy, try again: unconfigure failed |
Workaround: Download the SAN 4.3 packages from:
http://www.sun.com/storage/san/
The Sun Fire Link RSM (wrsm) driver is not loaded automatically at the kernel boot time if no wrsm configuration has been previously established.
When a DR unconfigure operation is performed on a system board containing permanent memory, all I/O devices are quiesced for approximately 75 seconds. The default timeout value for remote procedure calls to WRSM is 5 seconds, which means attempts to write to the WCI interrupt page during this quiescent period fail.
Workaround: Do not disconnect a system board containing permanent memory while a Sun Fire Link cluster is in use. If you do attempt to do so, you are prompted to confirm that you want to disconnect the board. Enter no to cancel the disconnect request:
# cfgadm -c disconnect N0.SB0 System may be temporarily suspended, proceed (yes/no)? no cfgadm: Configuration operation cancelled: disconnect N0.SB0 |
| Solaris 8 2/04 Release Notes Supplement for Sun Hardware | 817-4348-11 |
|
Copyright © 2004, Sun Microsystems, Inc. All rights reserved.