Skip Navigation Links | |
Exit Print View | |
Oracle Solaris 10 1/13 Release Notes Oracle Solaris 10 1/13 Information Library |
3. Oracle Solaris Runtime Issues
Recommended Adoption of SHA-256 and SHA-512 crypt Plug-ins for Password Encryption
System Might Encounter Data Integrity Issues (15825389)
System Reboots Continuously Because of a ZFS-Related Panic (15809921)
ZFS and UNIX/POSIX Compliance Issues
fdisk -E Command Can Sweep Disk Used by ZFS Without Warning (15325067)
BrightStor ARCserve Backup Client Agent for UNIX (Solaris) and ZFS Support Issue
ZFS GUI Should Check for /usr/lib/embedded_su Patch at the Beginning of Each Wizard (15287937)
Upgrading From Some Oracle Solaris 10 Releases Requires Remounting of File Systems
NFSv4 Access Control List Functions Might Work Incorrectly
Hardware-Related Issues and Bugs
SPARC: Dynamic Reconfiguration Operations Fail on an OPL Platform (15725179)
x64: PCI Subsystem ID Changes in ConnectX Firmware 2.6.0 From Mellanox (15544691)
The ZFS ARC Allocates Memory Inside the Kernel Cage Preventing DR (15377173)
Certain USB 2.0 Controllers Are Disabled
Supported USB Devices and Corresponding Hub Configurations
x86: Limitations Exist With Certain Device Drivers in the Oracle Solaris 10 OS
DVD-ROM/CD-ROM Drives on Headless Systems
x86: Manual Configuration Required to Specify Non-U.S. English Keyboards
Programs Compiled With Some Options of Oracle Solaris Studio Do Not Run (16237300)
Firefox and Thunderbird Hang in the Hangul Input Method (16043053)
Swedish Software Translations Note
Multiple Input Method Switcher Applications Appear in Trusted Java Desktop System
x86: Arabic Text Not Appearing in ar Locales
Migration Note for UTF-8 Locales
Emails Saved as Portable Format
File Names and Directory Names
Launching Legacy Locale Applications
Hardware for Some Keyboard Layouts Type 6 and 7 Not Available
System Domain of Interpretation Is Not Configurable (15283123)
IP Forwarding Is Disabled by Default in the Oracle Solaris 10 OS
Oracle Solaris Commands and Standards
Changed Man Pages for Trusted Extensions Are in Reference Manual Only
New ln Utility Requires -f Option
New tcsh Version Rejects setenv Variable Names That Use a Dash or an Equal Sign
STDIO getc Family EOF Condition Behavior Change
Output Columns of the ps Command Have Been Widened
Solaris Volume Manager Does Not Remove Devices Correctly If fdisk Does Not Have Valid Entries
Solaris Volume Manager metattach Command Might Fail
Email and Calendar Application
Problem With Changing Authentication Type (15256650)
User Preferences Not Fully Compatible
x86: Cannot Configure Full-Screen Magnification on Systems With One Video Card
System Administration Issues and Bugs
Oracle VTS fputest Fails on Fujitsu M10 Systems (15989998)
Oracle VTS testprobe_ramtest Fails on Fujitsu M10 Systems (15955560)
Oracle VTS testprobe_l3sramtest Fails on Fujitsu M10 Systems (15955589)
Crash Dump Is Saved in the /var/crash Directory (15767302)
Using DISM Might Cause the System to Hang (15807808)
x86: Running the ludelete Command Can Delete Shared Datasets Not Related To the OS (15817477)
Oracle Solaris 10 9/10 Clock Stops on Oracle VM 2.2 (15643194)
SPARC: FKU 137137-xx Patch Does Not Support Third-Party Volume Manager Software
Sun Patch Manager Tool 2.0 Incompatible With Previous Versions of the Tool
4. End-of-Software Support Statements
A. Previously Documented Bugs That Were Fixed in the Oracle Solaris 10 1/13 Release
This section describes system administration issues and bugs in this Oracle Solaris 10 OS.
fputest in the Oracle Validation Test Suite (Oracle VTS) checks whether a particular instruction is valid on a system before testing it. This check results in an incorrect value for the Fujitsu M10 systems causing the test to fail with the SIGILL signal.
testprobe_ramtest in Oracle VTS fails on Fujitsu M10 systems. An error message similar to the following is displayed:
ERROR: Test probe testprobe_ramtest() Failed with segmentation violation (signal: 11)
Workaround: The fix for this issue might be available on MOS in a later revision of the Oracle VTS 7.0 Patch Set 15.
testprobe_l3sramtest in Oracle VTS fails on Fujitsu M10 systems. An error message similar to the following is displayed:
ERROR: Test probe testprobe_l3sramtest() Failed with segmentation violation (signal: 10)
Workaround: The fix for this issue might be available on MOS in a later revision of the Oracle VTS 7.0 Patch Set 15.
Starting with the Oracle Solaris 10 1/13 release, by default, the crash dump file is saved in the /var/crash directory.
Note - Previously, the crash dump file was saved in the /var/crash/`uname -n' directory.
If you use the Dynamic Intimate Shared Memory (DISM) feature, there might be a performance regression and the system might hang depending on the size of the DISM used.
Workaround: Use one of the following workarounds depending on the architecture of your system.
For SPARC: Prevent the ISM and DISM from using a page size larger than 4 MB. Perform the following steps:
Add the following line in the /etc/system file.
set disable_ism_large_pages=0x74
Reboot the system.
For x86: Disable the support for a 1 GB page by setting the enable_1gpg variable to zero in kmdb. Perform the following steps:
Boot the system in kmdb mode.
Set the enable_1gpg variable to zero.
[0]> enable_1gpg/W 0 enable_1gpg: 0x1 = 0x0
Running the ludelete command can delete shared datasets that are not related to the OS in the following scenario.
Consider that a boot environment BE2 is created from BE1. After booting into BE2, you create a dataset dm2, and also create a boot environment BE3. Subsequently, if you change your boot environment to BE1 and choose to delete BE2, the dataset dm2 is also deleted.
The following example explains the scenario.
Consider that you have an active boot environment BE1 with an existing poolname that has various datasets containing data not related to the OS.
Create a boot environment BE2.
lucreate BE2
Activate BE2 and reboot.
luactivate BE2; init 6
Create a new zfs dataset dm2 under the existing poolname.
Create a new boot environment BE3.
lucreate BE3
Activate BE1 and reboot.
luactivate BE1; init 6
The active boot environment is now BE1.
Note - The poolname/dm2 is not automatically mounted as it was not originally a part of BE1.
Mount dm2.
mount poolname/dm2
Delete boot environment BE2.
ludelete BE2
The poolname/dm2 dataset is deleted.
Virtual machines (VMs) that are running Oracle Solaris 10 9/10 guests might face the following problems:
The time of day clock can stop moving forward.
Sleep can hang.
The guest VM can hang entirely.
Workaround: The Oracle Solaris guests must be pinned to physical CPUs. Pin the domain's virtual CPUs to the host's physical CPUs as follows:
# xm vcpu-pin domain vcpu cpus
See the xm(1) man page for more information.
The FKU 137137-xx patch does not support third-party volume manager software, with some exceptions. This lack of support is due to prepatch, postpatch, and postbackout implementation. If you use unsupported third-party volume manager software, you cannot apply the FKU patch. The following error message is displayed during patch installation:
unsupported root slice type xxxxx
Note that the Fujitsu and Veritas Volume Manager software is supported.
Workaround: None.
On systems that have an AHCI compliant SATA controller, the BIOS setup typically enables the controller to be set in AHCI, legacy, or RAID modes. The Oracle Solaris OS supports AHCI and legacy modes.
The SATA mode setting in BIOS must not be changed after an initial Oracle Solaris installation. The SATA mode setting must also not be changed before or after an Oracle Solaris upgrade. If the SATA mode BIOS setting is modified after the Oracle Solaris 10 OS is installed, the system will reset and fail to boot without indicating what led to the failure.
Workaround: If a boot failure occurs as a result of changing the BIOS setting, revert back to the original setting in order to boot the Oracle Solaris OS.
When run on large file systems, for example, ZFS, applications using the statvfs(2) or statfs(2) functions to get information about the state of the file system exhibit an error. The following error message is displayed:
Value too large for defined data type
Workaround: Applications should use the statvfs64() function instead.
A system that runs Sun Patch Manager 2.0 from Oracle can manage remote systems that run Patch Manager, including Sun Patch Manager 1.0.
However, a system with an earlier version of Patch Manager cannot manage remote systems that run Patch Manager 2.0. Earlier versions include the following:
Sun Patch Manager Base Software 1.x
Sun Patch Manager 1.0
Note - Common Information Model/Web Based Enterprise Management (CIM/WBEM) support for Patch Manager does not exist in the Solaris 8 OS. Consequently, remote management with Patch Manager does not apply to Solaris 8 systems.