This section describes the known OS provisioning (deployment) issues.
Perform the following steps to modify the default OS profile to provision Solaris 10 6/06 on a Sun Fire X4200 M2 server. This example uses an OS profile named s10u2_x86.
N1-ok> remove osprofile s10u2_x86 partition / N1-ok> remove osprofile s10u2_x86 partition swap N1-ok> add osprofile s10u2_x86 partition swap type swap size 1024 device c3t2d0s1 sizeoption fixed N1-ok> add osprofile s10u2_x86 partition / type ufs device c3t2d0s0 sizeoption free |
The deployment fails because the OS profile does not have the product key. However, attempting to add the product key to the OS profile fails with the message“OS profile is in use.”
Workaround: Wait until the job times out to add the product key to the OS profile.
When Red Hat OS is deployed on a system using the DHCP option, the system does not get an IP address from the DHCP range.
Workaround: Perform the following steps:
Log in to the management server.
In the /opt/sun/scs/data/allstart/scripts/dhcp_redhat.sh file, change the Device entry to the appropriate interface needed to run DHCP.
DEVICE=eth0 FILE=/etc/sysconfig/network-scripts/ifcfg-$DEVICE cat <<_EOF_> $FILE DEVICE=$DEVICE BOOTPROTO=dhcp _EOF_ chown root.root $FILE chmod 644 $FILE
Add the dhcp_redhat.sh script to the server's OS profile.
#n1sh add osprofile osprofile-name script /opt/sun/scs/data/allstart/scripts/dhcp_redhat.sh type=post
By default, the Sun Blade X8400 Server CMOS settings have ACPI enabled, which will cause Linux installation to go into interactive mode.
Workaround: Specify the additional kernel parameter to the load command.
N1-ok> load server servername osprofile profilename ... kernelparameter pci=nommconf
Specify this parameter each time you issue the load command.
If nfs is specified as the install protocol, OS deployment goes into interactive mode.
Workaround: Specify http as the install protocol.
The managed server fails to reboot completely after Red Hat Enterprise Linux 4.0AS Update 3, 64–bit operating system is provisioned on an Sun Blade X8400 server through the expansion module. The blade might fail to boot if Kudzu is enabled. The operation stops with the message Checking for new hardware.
To disable Kudzu on a Sun Blade X8400 Server through a Linux OS Profile, perform the following steps:
Log in to the management server as root.
Create a shell script file containing the following lines.
#!/bin/sh chkconfig kudzu off
Save the file in a known location, for example, /scripts/kudzu.sh.
Add the script to the post section of the OS profile from the N1 System Manager command line using the add osprofile command.
For example, if your Red Hat Enterprise Linux 4 Update 3 OS profile is named rhel4u3, you would then type:
N1-ok> add osprofile rhel4u3 script /scripts/kudzu.sh type=post |
Ensure that you add the script to the correct OS profile.
Redeploy the Red Hat Enterprise Linux 4 Update 3 to the Sun Blade X8400 Server using the updated profile.
Workaround: Deploy without specifying manual netboot.
When deploying an OS on a large number of managed servers, the OS deployments might fail due to the OS deployment jobs exceeding the job or job step timeout value. When this situation occurs, the job indicates the error but the error description is blank and the installation might have actually succeeded.
bash-3.00# n1sh show job 12 Job ID: 12 Date: 2006-03-29T01:06:53+0000 Type: Load OS Status: Error (2006-03-29T03:07:00+0000) Command: load group sparc-srvrs osprofile sol-sparc networktype=static ip=10.0.108.81-10.0.108.90 Owner: root Errors: 10 Warnings: 0 Steps ID Type Start Completion Result 1 Acquire Host 2006-03-29T01:06:54+0000 2006-03-29T01:06:54+0000 Completed 2 Acquire Host 2006-03-29T01:06:54+0000 2006-03-29T01:06:54+0000 Completed . . Servers 3 through 28 . 29 Acquire Host 2006-03-29T01:06:56+0000 2006-03-29T01:06:57+0000 Completed 30 Acquire Host 2006-03-29T01:06:56+0000 2006-03-29T01:06:57+0000 Completed 31 Execute Java 2006-03-29T01:06:56+0000 2006-03-29T03:06:58+0000 Error 1 32 Execute Java 2006-03-29T01:06:56+0000 2006-03-29T03:06:58+0000 Error 2 33 Execute Java 2006-03-29T01:06:56+0000 2006-03-29T03:06:58+0000 Error 3 34 Execute Java 2006-03-29T01:06:56+0000 2006-03-29T03:06:58+0000 Error 4 35 Execute Java 2006-03-29T01:06:56+0000 2006-03-29T03:06:58+0000 Error 5 36 Execute Java 2006-03-29T01:06:56+0000 2006-03-29T03:06:58+0000 Error 6 37 Execute Java 2006-03-29T01:06:57+0000 2006-03-29T03:06:58+0000 Error 7 38 Execute Java 2006-03-29T01:06:57+0000 2006-03-29T03:06:58+0000 Error 8 39 Execute Java 2006-03-29T01:06:57+0000 2006-03-29T03:06:58+0000 Error 9 40 Execute Java 2006-03-29T01:06:57+0000 2006-03-29T03:06:58+0000 Error 10 Errors Error 1: Description: Error 2: Description: . . Blank error messages 3 through 8 . Error 9: Description: Error 10: Description: |
Workaround: Increase the job timeout value using n1smconfig or try to reduce the reasons for exceeding the existing timeout value, for example, network latency problems. Also, some of the OS deployments might have succeeded. You can verify whether a server installed properly by viewing the OS Usage column in the All Servers page or by using the Serial Console to log in to the server.
The create os command will fail for Windows if the RIS server's host key does not exist in the management server's ./ssh/known_hosts file.
Workaround: Make sure the RIS server's host key exists in the management server's ./ssh/known_hosts file. Manually logging into the RIS server from the management server using the ssh command will automatically create a host key.
When you deploy a Windows OS with the networktype=dhcp attribute, the IP address assigned to the managed server after Windows is installed is the IP address specified by the bootip attribute. The IP address should be automatically assigned by the DHCP service.
Workaround: After the installation completes, reboot the managed server. The server's provisioning IP address will be correctly reassigned from the DHCP server.
Deploying Solaris 10 6/06 OS on a Sun Fire X4500 Server requires a special profile. You cannot choose any random disks. The system needs to be booted first either from the factory installed OS or from a CD-ROM to determine the server's correct boot disk.
To discover the correct boot disk, perform the following steps:
Open a terminal window.
Find the first bootable disk by typing
#cfgadm | grep sata3/0
You should see a message similar to the following example.
sata3/0::dsk/cXt0dY
Where X and Y are numbers, for example, c4t0d0.
To find the second bootable drive, perform either of the following actions:
Increment 0 by 4. For example, if the first drive is c4t0d0, then the second drive is c4t4d0.
Or type the command: #cfgadm | grep sata3/4
You should see a message similar to the following example.
sata3/4::dsk/cNtNdN
Where N is a number, for example, c4t1d0.
Proceed through the install using the information in Step 2 and Step 3 and apply it to the server's OS profile.
Workaround: No workaround exists.
As documented in the N1System Manager product documentation, the bge1 device path for Sun Fire V20z servers is /pci@0,0/pci1022,7450@a/pci17c2,10@3. However, when specifying the bge1 device path for the bootpath attribute, the OS deployments will fail on newer Sun Fire V20z servers because the bge1 device path has changed.
Workaround: On newer Sun Fire V20z servers, use the following bge1 device path for the bootpath attribute: /pci@0,0/pci1022,7450@a/pci17c2,10@2,1
The baud rate for the BIOS console must be set to 9600 (default) or OS deployment to a Sun Fire V20z or V40z server will fail. Therefore, you should not change the consolebaudrate value in the load server command or the Load OS wizard in the browser interface.
If the SP console baud rate is set to something other than 9600, the OS deployment will succeed but the console through the connect server command will display garbage characters.
Workaround: You must change the baud rate for the BIOS console manually after an OS deployment. Reboot the target server and enter the BIOS setup screen during the boot sequence. Consult the server's user manual to see how to change its BIOS settings.
When the user tries to load a OS profile on a server that is running the Grid Engine application, the operation fails with the message “Operation failed.”
Workaround: Unload the Grid Engine application from the target server.