Experiment No.

-6
Aim: - Learning of installation and upgradation of Linux operating
system.
Installing Linux on a personal computer may not be as difficult as you think. This document
explains how to install Linux on a PC, starting at the beginning: choosing a distribution
The first step for setting up Linux on a PC is the most time consuming, it is simply to use
a run from CD version of all distributions that you are interested in to pick the version you want
to install. Once you pick the distributions you want to check out, go to the distribution's Web
sites and download the live CD/move ISO image for it. Then, using the method for your CD
burning software, burn the ISO image to CDROM. With the CDROM in the drive, reboot the
computer and you are now running the Linux version.
Once you have selected which distribution(s) you will install, you can go back to that
distribution's Web site and download the install version of the distribution, which is most
commonly made up of three CDROM ISO images. After these are downloaded, burn the ISO
images to CDROM.
Delete the images from your hard drive after burning them to CDs.
You will then have to decide, before going any further, if you wish to multi-boot and
keep Windows on the system. If you are going to keep Windows, you will have to run a
defragmenting utility on the drive before going any further.
Completely cleaning up your hard drive, removing all temp files and defragmenting it
before backing up the data is always a good idea, and backing up your system is recommended
before installing a new operating system. Because Microsoft's NTFS is not 100 percent
supported, this is even more important if you are running any NT-based version of Windows.
If you have chosen to use a "Boxed set" of Linux, there is a set of "DOS" tools on one of
the CDs, which will run in Windows and allow you to resize the partition for Windows. If you
are using a downloaded version, then you can use the Linux tools during installation, but you
may have to completely delete the Windows partition. Getting a trial version of Partition magic
will allow you to resize the Windows partition without losing the data.
The partition table structure for a multi boot system looks complex, at first, but is actually
very simple:
 The first primary partition is your Windows partition.
 The first extended partition is a transfer partition for enabling read/write access to files
from all operating systems, and needs to be either a FAT32 partition or a FAT16 partition
 The second extended partition is the first Linux partition, and should be set up in the
Linux installation process. For only Windows and one Linux version, a 500MB partition
is more than enough room. It will be given the label of /boot [500MB in size ]
 You will need to create a partition with the label / [5 GB in size]
 Create a partition with the label /home [remaining amount of space for distribution]
  Create a partition with the label /usr [5 GB in size]
 Create a partition with the label /var [1 or 2 GB in size]
 Create a partition with the label /swap [double your RAM in size]
When creating the partitions you will notice that you have a number of file system type
options; the oldest one being the ext2fs. In a graphic installer, the file system types usually have
an explanation of what they mean. The current default file system is the reiserfs, a journaled file
system. This has a small hit on speed, but a major improvement on data protection, compared to
a non journaled file system.
Speed and reliability for data input/output being important, the reiserfs or the ext3fs are the
best supported file systems. If speed is more important than reliability, use the ext2fs. This does
not mean major risk of data loss, but power fluctuations may cause some data loss or corruption,
if you have spotty power, and do not have an interruptible power supply, go with a journalized
file system.
Follow the prompts for the distribution you are installing. Each distribution has a different
installation process, so detailing them here is not a viable option. If you are using one with a
console /text interface, the space bar selects, and the right arrow will expand a category, the left
arrow will collapse a category, up and down arrows will move you up and down in the listing.
The enter key will finish the selection process and start the installation. A graphic interface
installation program will have mouse support.

Selecting a package
Package selection is a big part of installing Linux. When first trying Linux it doesn't hurt
to install everything, as this enables you to see which applications you like and will use. There
are far more options than the live CD versions can include, so while you have some idea about
packages from having looked at the live CD, there are more options than those shown. Generally,
there are over 10 thousand packages in a three CD version of a distribution, including the
libraries for software development.
You may see a message screen after you have selected, or during the selection process,
that lists a number of packages. This is a list of packages required by the one you just selected,
called dependencies; you can just hit the ok button when you see it. Some installers allow for this
notification to be turned off, which I would recommend against, as the dependencies will help
you to see the different parts of each application.
During the installation you will be presented with a series of configuration options. The four
most important ones are:
 The firewall--you should turn it on, and at this time you can choose what traffic will be
allowed incoming from the Internet, if any.
 The Display settings--this will configure the graphic server for the GUI, and you should
use the test configuration option when doing this.
 The mouse--it is a good idea to specify which connection and type, as the any USB or
PS2 default of many distributions is not always a reliable mouse driver.
  The Internet connection--if you connect through a network card, then configure it as a
network connection, not a DSL connection, even if you use DSL.
If you tell Linux it's a DSL connection then Linux looks for the DSL modem as a device in the
computer and you will not get online. If you use dial up connection, and your modem is classed
as a Winmodem, then you will need to get the drivers for it from the linmodems Web site before
installing Linux. After installing Linux, you will have to compile the drivers for the modem,
install them and run the network configuration tool for the distribution you have installed.

Bootloader:-When the bootloader is installing, you have the option of setting which
bootloader to use, GRUB or LILO, either is a stable option. You can also set which OS will be
the default OS at boot. The screens will have a default box on them, if you wish to keep
Windows as default then just select default when highlighting Windows (note, it may be DOS
with some distributions). You can, if you want, delete the failsafe and nonframebuffer boot
options, keeping only Linux, Windows and floppy options. (Note, the floppy option is not always
included with newer distributions).
If you are installing several versions of Linux, the boot partition created during the first
Linux installation will need to be selected and labelled as /boot for each one. Make the same
partitions for each version, except for swap, that you only need one. Do NOT use any other pre-
existing Linux partition with different distributions, as the versions of software between the
distros may cause conflicts when booting into them. By using the same boot partition, the
bootloader installation step will read the data in it and configure the multi-boot for each new
distribution. It may be useful to label each distro by its name in the boot menu as you go, this
will make the boot options clearer when you are choosing one.
When updating the Red Hat EL or SLES operating system on nodes, if the Mode attribute
is PreManaged, then after the operating system upgrade, the Mode attribute value is changed to
Managed. If the Mode attribute is Managed, then it is not changed. If the Mode attribute is
MinManaged before the operating system upgrade, then the Mode attribute value remains
MinManaged.
If the installnode command fails for a PreManaged node, the attribute value of the node
is PreManaged or Installing. If the installnode command fails for a MinManaged node, the
attribute value of the node is MinManaged or MinManaged-Installing.
To upgrade the operating system and CSM on all the nodes, issue the following
command:
installnode -a
The Red Hat EL nodes with the InstallMethod attribute set to kickstart-upgrade are
rebooted and upgraded with the new level of the operating system, rebooted, and then CSM is
updated if necessary. The Red Hat EL operating system upgrade runs asynchronously.
Immediately after the operating system upgrade process is initiated (that is, when the node is
rebooted), the installnode command exits, even though the operating system upgrade is not
complete.
The SLES nodes with the InstallMethod attribute set to you are updated with the new
level of the operating system, rebooted to pick up the new kernel, and then CSM is upgraded if
necessary. The installnode command continues to run as the SLES operating system is updated.
This may take a while to run. Then, installnode reboots the node, and exits. The rest of the
upgrade process (CFM, SMS, osupgradepostreboot scripts, upgrading CSM) continues
asynchronously.
You can use the -t flag on the installnode command to provide a timeout value in
minutes. If you do not specify a value for timeout, the default is 60 minutes:
installnode -P -t timeout
If the operating system upgrade process does not complete within the timeout period specified,
CSM considers the operating system upgrade process as failed. You can use the monitorinstall
command to provide output information for the installation process.
After the operating system is updated, the following jobs run on the node:
1. The osupgradeprereboot customization scripts are run.
2. The node reboots to its local hard disk.
3. CSM is installed, along with the software listed in Planning for CSM for Linux.
4. The node's Mode attribute changes to Managed, or remains MinManaged.
5. SSH or RSH are set up on the node so that the node is accessible from the management
server.
6. Any CFM files are transferred to the node.
7. SMS is run to install or update software, if it is configured.
8. The osupgradepostreboot customization scripts are run.
If you defined Kerberos options with the csmconfig command when you defined the
management server, the installnode command sets up the Kerberos options for the cluster..
After an operating system upgrade the node BIOS boot order can remain:
1. diskette
2. CD-ROM
3. network
4. hard disk
Every time the node boots, it uses Dynamic Host Configuration Protocol (DHCP) to contact the
management server or install server, which uses pxelinux to boot the node from its hard drive.
Alternately, after the operating system upgrade is complete, you can change the boot order in the
BIOS to the following:
1. diskette
2. CD-ROM
3. hard disk
4. network