HP Insight Control for Linux 6.0 User Guide

IMPORTANT: You must ensure that the captured image kernel supports the selected file system
types, including the Logical Volume Manager (LVM). It is especially important that the initial
RAM disk of the captured image (initrd*.img) has file system support for the selected /boot
and / (root) partitions, because the initial RAM disks are not remade by the Deploy→Operating
System→Deploy Linux Image... task. It is possible to remake the initial RAM disk as part of a
post-installation script.
HP recommends that you follow these guidelines before you capture the Linux image from a
managed system:
• Do not capture the image on a system that has sensitive information on it.
• Before capturing an image from a Xen virtual host, stop all its virtual guests. For information
on the virtual machine operations, see Insight Control Virtual Machine Management User Guide.
• Set up a suitable disk partitioning scheme on the source system during the initial OS
installation and select the Use partition scheme from image option when you deploy the
captured image.
• Ensure that the system clock on the CMS and the managed systems are synchronized before
you capture or deploy a Linux image. For more information about how unsynchronized
system clocks can affect an image deployment, see Section 1.8 (page 17).
• When deploying a captured image to a managed system other than the one from which it
was originally captured, HP recommends ensuring that your source node obtains its IP
address and host name from DHCP instead of hard-coding those values because Insight
Control for Linux automatically modifies the network configuration of the new managed
system to prevent address and name collisions. These modifications include:
— Removing or commenting any hard-coded host name references.
— Disabling any network interfaces that were assigned a static IP address.
— Replacing the source system's MAC addresses with the target system's MAC addresses
in relevant configuration files.
If you must use static addresses and host names, you need to create a Postdeployment script
capable of setting these values.
• For SLES images, change the hard links to soft links before capturing the image.
SLES relies on the use of hard links within its file system, and those hard links are captured
by the tar command used to capture the image.
If a partitioning scheme is used during deployment that distributes files to multiple file
systems (like separate /usr and /var partitions), the tar command does not allow hard
links to be established across separate file systems. This generates an error, causing the task
to fail.
• Follow these guidelines when you capture a Linux OS from a managed system with multiple
disks:
— Each disk you want to capture must have at least one partition. Disks mounted with
no partitions are not supported.
— You must modify the /etc/fstab file to include the mount points of all partitions on
the disks you want to capture.
Because all disks are repartitioned during an image deployment operation, you must
ensure that the contents of the disks you want to capture are listed correctly in the
/etc/fstab file.
For a capture operation to be successful, the dump field for each partition to be captured
must be set to 1. The dump field is the fifth field in the standard mount line. Setting
this field to 1 enables the partition to be captured, and setting this field to 0 (zero)
bypasses the capture.
10.2 Prerequisites to capturing a Linux image 113