The following bugs might occur during or after the installation of Solaris 10 OS.
For GUI-based installation of the Solaris 10 10/09 OS, the system should have a minimum 768 MB of memory. If the physical memory is less than 786 MB, an error message similar to the following is displayed and the installation switches to text installation.
Not enough memory for graphical installation. Graphical installation requires 768 MB of memory. Found 640 MB of memory. Reverting to text-based installation. |
Workaround: Ensure that you have 768 MB of memory for a GUI-based installation or upgrade.
Disk space reallocation (DSR) Upgrade using DVD or /net image might cause corruption of systems. Upgrade appears successful on the system console, but no packages are added after patch and package removal. After removal of patches or packages, the upgrade goes directly to creating boot blocks and reboots the system. As all packages, including ones that are necessary for booting, are removed and no packages are added, the system fails to boot. This failure is seen intermittently.
The following error message is displayed:
Can't open boot_archive Fast Data Access MMU Miss %TL:1 %TT:68 %TPC:f0009ca4 %TnPC:f0009ca8 %TSTATE:8840001600 %PSTATE:16 ( IE:1 PRIV:1 PEF:1 ) DTLB-MATCH:ffffffffffffe ( VA:-2000 CTX:0 ) |
Workaround: Use Solaris Live Upgrade in place of DSR upgrade for change disk allocation.
On systems with low memory, installation might fail if either / or /var have just minimum or recommended disk sizes. The following error message is displayed:
Creating boot_archive for /a updating /a/platform/sun4u/boot_archive 15+0 records in 15+0 records out cat: write error: No space left on device |
Workaround: Increase the size of /var to at least 300 Mbytes on low memory systems.
DSR upgrade fails on systems installed with a Solaris 10 release prior to the Solaris 10 10/09 release. This failure is due to an auto-layout problem. For example, if you install with the following profile:
install_type initial_install system_type standalone cluster cxall partitioning explicit filesys rootdisk.s0 auto / logging filesys rootdisk.s1 512 swap filesys rootdisk.s6 10240 /backup logging filesys rootdisk.s7 free |
And, then you do DSR upgrade to the Solaris 10 10/09 release with the following profile:
install_type upgrade root_device rootdisk.s0 backup_media local_filesystem /backup layout_constraint rootdisk.s0 changeable layout_constraint rootdisk.s7 available |
The DSR upgrade fails with the following error message:
Auto-layout could not determine how to reallocate space on the file systems with the constraints you specified. Try other constraints |
Workaround: Label the rootdisk.s1 and rootdisk.s6 slices as “changeable”. If an extra disk is available in the system, and there is sufficient space on an unused slice for a backup, then modify the <backup_media> keyword, as shown in the following example:
backup_media local_filesystem /dev/dsk/c0t1d0s7 |
In this example, disk c0t1d0s7 is the second disk available, and the disk has enough space for a backup. Alternately, if there is only one rootdisk in the system, perform the backup on a remote system. Use the following syntax:
backup_media remote_filesystem hostname:/export/backup |
The DSR upgrade will succeed.
On multi-socket Nehalem systems, where DIMMs are installed on only one CPU, the system panics during boot with the following kernel panic error message:
Too many lgrps for platform (4) |
Workaround: Disable the use of the ACPI System Resource Affinity Table (SRAT) by performing one of the following two options:
Boot into kmdb(1) by adding '-k' to the GRUB menu kernel line, booting, and typing the following commands to kmdb(1):
lgrp_plat_srat_enable/W 0 :c |
When the system is booted, set the <lgrp_srat_enable> boot property to 0 using eeprom(1M). This setting disables the use of SRAT across reboots.
Localized Solaris installer always runs in the EUC or ISO8859 locales. Thus, the install log is written in EUC or ISO8859 encoding.
Workaround: While performing localized CLI (command line interface) install through a terminal, the locale on the terminal must be EUC or ISO8859 locale.
The installation fails, when installing WAN boot image from Solaris 10 10/09 release. The following error message is displayed:
boot_file=/wanboot/wanboot.u-137111-04 root_file=/miniroot/miniroot.s10-1008 |
Workaround: The Solaris 10 10/09 miniroot image must be patched to proceed with the installation. Perform the following steps to patch a network installation miniroot image:
These steps assume that you have a system on your network that is running the current Solaris release, and that system is accessible over the network.
On a system that is running the current Solaris release, log in as superuser or assume an equivalent role.
Change to the Tools directory of the installation image you created.
# cd install-server-path/install-dir-path/Solaris_10/Tools |
Specifies the path to the install server system on your network, for example, /net/installserver-1.
Create a new installation image, and place that image on the system that is running the current Solaris release.
# ./setup_install_server remote_install_dir_path |
Specifies the path on the current Solaris release in which to create the new installation image.
This command creates a new installation image on the current Solaris release. In order to patch this image, you must temporarily place this image on a system that is running the current Solaris release.
On the current Solaris release, unpack the network installation boot archive.
# /boot/solaris/bin/root_archive unpackmedia remote_install_dir_path \ destination_dir |
Specifies the path to the network installation image on the current Solaris release.
Specifies the path to the directory to contain the unpacked boot archive.
On the current Solaris release, patch the unpacked boot archive.
# patchadd -C destination_dir path-to-patch/patch-id |
Specifies the path to the patch that you want to add, for example, /var/sadm/spool.
Specifies the patch ID that you want to apply.
You can specify multiple patches with the patchadd -M option. For more information, see patchadd(1M).
Don't use the patchadd -C command unless you have read the Patch README instructions or have contacted your local Sun support office.
On the current Solaris release, pack the boot archive.
# /boot/solaris/bin/root_archive packmedia remote_install_dir_path \ destination_dir |
Copy the patched archives to the installation image on the install server.
# cd remote_install_dir_path # find boot Solaris_10/Tools/Boot | cpio -pdum \ install-server-path/install_dir_path |
If you are installing a Solaris Flash archive from a release before the Solaris 10 10/08 release, the archive fails to install. Starting with the Solaris 10 10/08 release, if the archive is from the Solaris 10 10/08 release or later, the archive can be installed. The following error message is displayed:
bootadm: biodev command failed for disk: /dev/dsk/<c0t2d0s0>. bootadm: 1s_bootdisk(): cannot determine BIOS disk ID "hd?" for disk: /dev/dsk/<c0t2d0s0> bootamd: get_grubroot(): cannot get (hd?, ?,?) for menu. menu not on bootdisk: /dev/rdsk/<c0t2d0s0> |
Workaround: Choose one of the following workarounds:
Install a Solaris Flash archive from the Solaris 10 10/08 release or later.
If you need to install a Solaris Flash archive from a previous release, boot from the previous release and extract the archive.
Installation fails if you attempt to boot the system by using the following command:
# boot -F wanboot -vV -o prompt |
As a result, a system panic occurs. The following error message is displayed:
wanboot: panic - boot: can't alloc ramdisk memory |
This problem occurs on both Sun4u and Sun4v platform types. This problem was introduced in the Solaris 10 10/08 release.
Workaround: Do not use wanboot program to boot the system. Instead, boot the system directly from the installation media.
Some Asian locales like, th_TH.ISO8859-11, th_TH.TIS620, ko_KR.EUC, ko_KR.UTF-8, zh_TW.EUC, zh_CN.EUC, zh_CN.GBK, and zh_CN.UTF-8 cannot be used while installing with Custom JumpStart. While setting locales in the sysidcfg file using the system_locale keyword, the following error message is displayed:
xx_xx.xxxxx is not a valid system locale |
The JumpStart installation is stopped and Interactive installation starts.
Workaround: Use shorter locale names such as th_TH, ko, ko.UTF-8, zh_TW, zh, zh.GBK, and zh.UTF-8, instead of longer locale names,
While upgrading Solaris Trusted Extensions from the Solaris 10 11/06 or Solaris 10 8/07 release to the Solaris 10 10/08, Solaris 10 5/09 release or the current Solaris 10 10/09 release, the prodreg entry for Solaris Trusted extensions is not removed. No error message is displayed.
Workaround: After upgrading Solaris Trusted Extensions to the current release, remove the prodreg entry manually as follows:
# prodreg unregister -f -r -u "Solaris Trusted Extensions" -i 1 |
The driver for the NVidia SATA controller MCP55 and CK804 chipsets included in the Solaris 10 10/09 release does not support ATAPI CD/DVD devices.
During system installation or upgrade by using a SATA CD/DVD device, the Solaris installer starts and appears to function but fails with the following error messages:
ERROR: The disc you inserted is not a Solaris OS CD/DVD |
Or
There were problems loading the media from /cdrom |
During normal system operation, the SATA CD/DVD drive is not configured and cannot be used.
Workaround: Use the USB CD/DVD.
During a Solaris upgrade, if you select Detailed Analysis to see the patches that will be removed, the panel that displays the patches is not scrollable. The complete list of patches to be removed cannot be viewed.
Workaround: Run the analyze_patches scripts manually:
# cd <cdrom>/Solaris_10/Misc # ./analyze_patches -R rootdir -N netdir -D databasedir |
The command options are as follows:
rootdir is the root of the installed system. The default root directory is /.
netdir is the path to the root of the OS image to be installed. /cdrom/cdrom0 is the default path. netdir is the path to the directory that contains the Solaris_10_606 directory. You must use this option if you are running the patch_analyzer from an NFS mount point.
If the script is called from a directory other than the /Misc directory in the OS image, the program cannot find the database it uses for patch analysis. Use the -D option to supply the path to the database. Without this database, which is located in the Solaris_10_606/Misc/database directory on the OS image, the script will not run correctly.
During a Solaris Serial installation, the system does not have an attached keyboard, but a prompt for selecting the keyboard layout is displayed.
Workaround: Select the keyboard layout. You can change the keyboard layout using the kbd command with -s option if needed when a keyboard is attached.
The Solaris OS might panic with a null pointer dereference if the driver.conf file has been modified using the parent attribute. The following error message is displayed:
System panic: BAD TRAP: type=31 rp=2a101d31200 addr=8 mmu_fsr=0 occurred in module "pcie" due to a NULL pointer dereference |
Workaround: Before you install the Solaris 10 10/09 OS on a PCI Express (PCIe) based SPARC system, check if you have modified the driver.conf files as described below:
Change to the /kernel/drv directory.
Check if the parent attribute is set in the driver.conf file. If the parent attribute has been set and the last node in the parent has the word pci in it, comment out that line. Ensure that you comment out the entire property. For example:
# name="ACME,simple" parent="/pci@7c0/pci@0/pci@9" # unit-address="3,1" debug-mode=12; name="ACME,example" parent="pseudo" instance=1; name="ACME,scsi" parent="/pci@7c0/pci@0/pci@9/scsi" unit-address="3,1" debug-mode=12; |
Repeat step 2 for the driver.conf files in the following directories:
/usr/kernel/drv
/platform/`uname -i`/drv
/platform/`uname -m`/drv
Install the Solaris 10 10/09 OS.
Install patch 127747-01.
Change to the directories listed in steps 1 and 3. Uncomment all the lines that were commented out in step 2 in the driver.conf files.
If Linux is installed on your disk and you installed the Solaris OS on a separate partition, the Linux partition does not display on the GRUB menu. No error message is displayed.
Workaround: Edit the GRUB menu's menu.lst file to add Linux to the GRUB menu. Perform the following steps:
Boot the Solaris OS.
Edit the menu.lst file at /boot/grub/menu.lst. For more information, see the System Administration Guide: Basic Administration.
Installations can run out of memory and hang on 512-Mbyte systems under the following conditions:
You install using a network installation image of the OS.
The Solaris installation program is in a windowing environment:
The graphical user interface (GUI).
The -text option or “3. Solaris Interactive Text (Desktop session)” in the menu below.
When the system exhausts available memory, the GUI installation process slows and eventually fails. The text does not display in a windowing environment.
Workaround:
To avoid this problem, select a non-windowing environment during the initial installation startup. During startup from the installation media the following menu is displayed:
----------- 1. Solaris Interactive (default) 2. Custom JumpStart 3. Solaris Interactive Text (Desktop session) 4. Solaris Interactive Text (Console session) 5. Apply driver updates 6. Single user shell Enter the number of your choice. ----------- |
At this point, choose “4. Solaris Interactive Text (Console session)”. This non-windowing environment will initiate a text-only installation, without using memory-intense processes.
If you install the Solaris 10 10/09 OS on an x86 based system, the following error message is displayed.
/sbin/dhcpinfo: primary interface requested but no primary interface is set |
The error does not affect the installation, and the installation succeeds.
Workaround: Ignore the error message.
A problem might occur when you are using the Solaris installation program and are using CD media. After the installation of the Solaris software from the Solaris 10 10/09 Software CD - 4, the following prompt is displayed:
Press Reboot now to continue. |
If you press Reboot, the system might not respond. The installation is successful. However, the error prevents the installation program from exiting normally. Consequently, the typical postinstallation cleanup and system reboot cannot occur.
An error message similar to the following example is logged in /tmp/disk0_install.log file:
Exception in thread "Thread-70" java.lang.IndexOutOfBoundsException: Index: 6, Size: 5 at java.util.ArrayList.add(ArrayList.java:369) at com.sun.wizards.core.WizardTreeManager. actualExitButtonPressed(WizardTreeManager.java:1499) at com.sun.wizards.core.WizardTreeManager. exitButtonPressed(WizardTreeManager.java:1486) at com.sun.wizards.core.AutonextController. run(AutonextController.java:736) at java.lang.Thread.run(Thread.java:595) |
Workaround: Choose one of the following workarounds.
Workaround 1: Follow these steps.
Shut down the system, then turn on the power again.
After the system starts up, insert the appropriate CD at the prompt. Although the CD is inserted, no additional software is installed on the system. In a moment, the system reboots normally.
Workaround 2: Follow these steps.
Open a terminal window.
Log in as superuser.
Create or modify access to the .instsuccess file.
# touch /tmp/.instsuccess |
Stop the Java process.
# pkill -9 java |
The system reboots without prompting for the CD media.
If you use the Custom JumpStart installation method to install on an x86 based system, and explicitly configure slice 2 as the overlap slice in the profile, an error occurs. The system does not reboot successfully after the installation is completed. The following error message is displayed:
Cannot find Solaris partition |
This failure occurs because the overlap slice 2 (c0t0d0s2, for example) is set to begin at cylinder 1 rather than cylinder 0.
Workaround: In the Custom JumpStart profile, remove the filesys keyword entry that configures slice 2 as the overlap slice. For example, you would remove a keyword entry that is similar to the following entry.
filesys c0t0d0s2 all overlap |
After you remove the entry, perform the Custom JumpStart installation.