Tài liệu Advanced Installation docx

Advanced Installation CERTIFICATION OBJECTIVES 5.01 RAID Configuration 5.02 Using Kickstart to Automate Installation 5.03 Understanding Kernel Modules 5.04 The /lib/modules/kernel_versio

Advanced Installation

CERTIFICATION OBJECTIVES

5.01 RAID Configuration 5.02 Using Kickstart to Automate

Installation 5.03 Understanding Kernel Modules 5.04 The /lib/modules/kernel_version/

Directory Structure

✓ Two-Minute Drill

Q&A Self Test

In this chapter, you will learn how to manage Linux in advanced installation and configuration

scenarios The installation topics cover creating automated installation scripts and managing

kernel modules You will learn how to implement a Redundant Array of Inexpensive Disks

(RAID), as well as an array of disks for the Logical Volume Manager (LVM), and master the intricate

details of the automated Kickstart installation process Finally, you’ll get a basic sense of how you

can modularize the kernel to your advantage

Remember, one of the three RHCE exams is based on how well you know theinstallation process By the time you finish this chapter, you should be ready to installLinux in an automated fashion from a local boot disk or over a network from an NFS

or HTTP server And as you work with kernel modules near the end of the chapter,you’ll examine some of the techniques you can use on the RHCE troubleshootingexam to ensure that the kernel is properly set up to work with your hardware

physical failure on one of the disks Alternatively, you can configure software-based

RAID on multiple partitions on the same physical disk While this can protect youfrom a failure on a specific hard drive sector, it does not protect your data if there is

a failure of the entire physical hard drive

Depending on your definitions, RAID has nine or ten different levels, which canaccommodate different levels of data redundancy Only three levels of RAID aresupported directly by current versions of Red Hat Linux: levels 0, 1, and 5 HardwareRAID uses a RAID controller connected to an array of several hard disks A driver

RAID Configuration 251

must be installed to be able to use the controller Linux, meanwhile, offers a softwaresolution to RAID with the md kernel module Once RAID is configured on yoursystem, Linux can use it just as it would any other block device

The RAID md device is a meta device In other words, it is a composite of two

or more other devices such as /dev/hda1 and /dev/hdb1 that might be components

of a RAID array.

The following are the basic RAID levels supported by Red Hat Linux In addition,Red Hat Linux is starting to incorporate the Logical Volume Management (LVM)system Theoretically, it will allow you to resize or reallocate partitions as your needsevolve In practice, LVM is new to Red Hat, and support for this system is not complete

RAID 4

While this level of RAID is not directly supported by current versions of Red HatLinux, it is still supported by the current Linux kernel RAID 4 requires three or

more disks As with RAID 0, data reads and writes are done in parallel to all disks.

One of the disks maintains the parity information, which can be used to reconstructthe data Reliability is improved, but since parity information is updated with everywrite operation, the parity disk can be a bottleneck on the system RAID 4 is known

as disk striping with parity

RAID 5

Like RAID 4, RAID 5 requires three or more disks Unlike RAID 4, RAID 5 distributes,

or “stripes,” parity information evenly across all the disks If one disk fails, the data can

be reconstructed from the parity data on the remaining disks RAID does not stop;

all data is still available even after a single disk failure RAID level 5 is the preferredchoice in most cases: the performance is good, data integrity is ensured, and only onedisk’s worth of space is lost to parity data RAID 5 is also known as disk striping withparity

Hardware RAID systems should be “hot-swappable.” In other words, if one disk fails, the administrator can replace the failed disk while the server is still running The system will then automatically rebuild the data onto the new disk Since you can configure different partitions from the same physical disk for a software RAID system, the resulting configuration can easily fail if you use two or more partitions on the same physical disk.

The exam may use examples from any level of RAID.

RAID in Practice

RAID is associated with a substantial amount of data on a server It’s not uncommon

to have a couple dozen hard disks working together in a RAID array That much datacan be rather valuable

If continued performance through a hardware failure is important, you can assignadditional disks for “failover,” which sets up spare disks for the RAID array When onedisk fails, it is marked as bad The data is almost immediately reconstructed on thefirst spare disk, resulting in little or no downtime The next example demonstratesthis practice in both RAID 1 and RAID 5 arrays Assuming your server has four drives,with the OS loaded on the first, it should look something like this:

Ill 5-1

All four drives (hda, hdb, hdc, hdd) should be approximately the same size.

This first example shows how to mirror both the /home and the /var directories(RAID 1) on Drive 2 and Drive 3, leaving Drive 4 as a spare

You need to create nearly identically sized partitions on Drives 2 and 3 In thisexample, four disks are configured with four partitions of the same size Mark thelast two partitions on all drives as type 0xFD (for autodetection) using the Linuxfdisk program You can use the “t” option to toggle the drive ID type

In the partition table of the first drive is /dev/hda3 (currently mounted as /home)and /dev/hda4 (currently mounted as /var) The second drive includes /dev/hdb3and /dev/hdb4 The third drive is set up with /dev/hdc3 and /dev/hdc4, while thelast drive has /dev/hdd3 and /dev/hdd4 All of these partitions have been markedwith partition IDs of type 0xFD

Next, update the configuration file /etc/raidtab as follows:

raiddev /dev/md0 raid-level 1 nr-raid-disks 3 nr-spare-disks 1 persistent-superblock 1 chunk-size 4

device /dev/hda3 raid-disk 0 device /dev/hdb3 raid-disk 1

RAID Configuration 253

device /dev/hdc3 raid-disk 2 device /dev/hdd3 spare-disk 0

raiddev /dev/md1 raid-level 1 nr-raid-disks 3 nr-spare-disks 1 persistent-superblock 1 chunk-size 4

device /dev/hda4 raid-disk 0 device /dev/hdb4 raid-disk 1 device /dev/hdc4 raid-disk 2 device /dev/hdd4 spare-disk 0

Table 5-1 shows what some of the commands are, along with a brief description

of what they do

Take special note that raid-disks and spare-disks start counting at 0; nr-raid-disks and nr-spare-disks are the correct number of drives For example: If nr-raid-disks

= 3, then the raid-disks are 0, 1, and 2.

If necessary, now is the time to convert these partitions to the default ext3filesystem You can do this by adding journaling to each partition Based on theprevious example, the commands would be:

tune2fs -j /dev/hda4 tune2fs -j /dev/hdb4 tune2fs -j /dev/hdc4 tune2fs -j /dev/hdd4

Alternatively, you could have set up an ext3 filesystem when you configured a specific

partition; such as with the mkfs -j /dev/hda4 command The -j switch sets up a journal.

When a journal is added to the ext2 filesystem, it creates an ext3 filesystem

Now to start RAID 1, initialize the md0 and md1 devices with the followingcommands:

mkraid /dev/md0; mkraid /dev/md1

parity-algorithm right-symmetric device /dev/hda4

raid-disk 0 device /dev/hdb4 raid-disk 1 device /dev/hdc4 raid-disk 2 device /dev/hdd4 spare-disk 0

Again, run mkraid /dev/md0 to initialize RAID 5.

Formatting the RAID Partition

Before you run mke2fs to format each partition, you should understand how to usethe special mke2fs stripe option For instance, if you have a chunk-size of 32KB, use

64 blocks per chunk If you format using 4K block sizes, then use 8 blocks per chunk

If you specify the chunk size (stride) when you format a RAID 5 device, you will see

a considerable increase in performance For example, the following command sets up

a format with a 4K block size and 8 blocks per chunk on RAID device /dev/md0

nr-raid-disks Number of RAID disks to use nr-spare-disks Number of spare disks to use persistent-superblock Required for autodetection chunk-size Amount of data to read/write parity-algorithm How RAID 5 should use parity

TABLE 5-1

RAID Commands

The mkfs command is a “front-end” to mke2fs for most standard Linux partitions.

In other words, both commands work the same for most Linux partitions.

For autodetection to work properly, you need to have the partitions set to type0xFD, as described earlier You also must have autodetection turned on in the kernel,and use the persistent-superblock option If all is well, when the kernel boots, it willautomatically detect RAID and fix any errors from crashes during the boot process.Setting up a RAID level root device is a bit trickier Because of the importance ofthis data, manually copy the contents of the root partition to a different drive Set upand write /etc/lilo.conf or /etc/grub.conf to the other drive as well

Then if there’s a problem creating the root RAID array, you can still boot Linuxfrom the other drive Fortunately, this is an advanced skill beyond the scope of theRHCE exam For more information on this process, go to the Linux DocumentationProject Web site and read the Root-RAID-HOWTO

EXERCISE 5-1

Mirror the /home Partition Using Software RAID

Don’t do this exercise on a production computer If you have a computer with RedHat Linux already installed that you can use for testing, that is best One alternative

is to use virtual machine technology such as VMWare, which can allow you to set

up these exercises with minimal risk to a production system When you’re ready, usethe Linux fdisk techniques discussed in Chapter 3 to configure the following two-drivepartition scheme:

Drive 1:

hda1 256 / hda2 64 swap hda3 500 /home hda4 256 /var Drive 2:

hdb1 1200 /usr hdb2 64 swap hdb3 100 /tmp hdb4 500 (not allocated)

Now create a mirror of hda3 to hdb4 partition (The partition sizes do not have to

be identical.)

RAID Configuration 257

If you’re making fdisk changes on a production computer, back up the /home partition first Otherwise, all data on the current /dev/hda3 will be lost.

1 Mark the two partition IDs as type 0xFD using the Linux fdisk utility

[root]# fdisk /dev/hda Command (m for help) : t Partition number (1-4) 3

Partition ID (L to list options): FD Command (m for help) : w

Command (m for help) : q

# fdisk /dev/hdb Command (m for help) : t Partition number (1-4) 4

Partition ID (L to list options): FD Command (m for help) : w

Command (m for help) :q

2 Update the configuration file /etc/raidtab with these lines of code:

[root]# vi /etc/raidtab raiddev /dev/md0

raid-level 1 nr-raid-disks 2 nr-spare-disks 0 persistent-superblock 1 chunk-size 4

device /dev/hda3 raid-disk 0 device /dev/hdb4 raid-disk 1

3 Now make the RAID device file md0 and format it this way:

[root]# mkraid /dev/md0 [root]# mke2fs -b 4096 -R stride=8 /dev/md0

4 Update each device to the ext3 filesystem by adding journaling

(Alternatively, you could have just added the -j switch when using mke2fs toformat the /dev/md0 device.):

tune2fs -j /dev/hda3 tune2fs -j /dev/hdb4

Logical Volume Management

Logical Volume Management (LVM) should allow you to manage active partitions.For example, if you find that you have extra space on the /home directory partitionand need more space on your /var directory partition for log files, LVM will let youreallocate the space Alternatively, if you are managing a server on a growing network,new users will be common You may come to the point where you need more room

on your /home directory partition With LVM, you can add a new physical disk, andallocate its storage capacity to an existing /home directory partition

As LVM is new to Red Hat Linux 8.0, it seems unlikely that Red Hat will make thispart of the RHCE 8.0 exam Nevertheless, LVM is potentially too good of a tool forany Linux administrator Other Linux distributions such as S.u.S.E have incorporatedLVM in their distributions for some time now I would expect Red Hat to incorporateLVM requirements in later versions of the RHCE exam In either case, check theRHCE exam guide for the latest information

While LVM can be an important tool to manage partitions, it does not by itself provide redundancy Do not use it as a substitute for RAID However, you can use LVM in addition to a properly configured RAID array.

LVM Concepts

The LVM system organizes hard disks into Logical Volume (LV) groups Essentially,physical hard disk partitions (or possibly RAID arrays) are set up in a bunch ofequal-sized chunks known as Physical Extents As there are several other conceptsassociated with the LVM system, start with some basic definitions:

■ Physical Volume (PV) is the standard partition that you add to the LVM mix.

Normally, a physical volume is a standard primary or logical partition It canalso be a RAID array

■ Physical Extent (PE) is a chunk of disk space Every PV is divided into a number

of equal sized PEs Every PE in a LV group is the same size Different LV groupscan have different sized PEs

■ Logical Extent (LE) is also a chunk of disk space Every LE is mapped to

a specific PE

■ Logical Volume (LV) is composed of a group of LEs You can mount

a filesystem such as /home and /var on a LV

■ Volume Group (VG) is composed of a group of LVs It is the organizational

group for LVM Most of the commands that you’ll use apply to a specific VG

The first step to create a LVM is to initialize a physical disk To create a new PV

on a freshly installed hard disk such as /dev/hdc, run the following command:

# pvcreate /dev/hdc

This command won’t work if you already have set up a partition table on this drive.Alternatively, you can set up a new PV on a properly formatted partition For example,assume that you’ve added a new partition, /dev/hdc1 You could then use fdisk or

a similar tool to set it to the Linux LVM partition type In fdisk, this corresponds topartition type 8e Once your partition is ready, you can run a command such as thefollowing to create a new PV:

# pvcreate /dev/hdc1

Once you have two or more PVs, you can create a Volume Group (VG) You can

substitute the name of your choice for volumegroup.

# vgcreate volumegroup /dev/hdc1 /dev/hdd1

When you create a new partition that you want to add to the VG, just extend it withthe following command:

# vgextend volumegroup /dev/sda1

But a new VG doesn’t help you unless you can mount a filesystem on it So you need

to create a logical volume (LV), for this purpose The following command creates a LV.You can add as many chunks of disk space (aka Physical Extents - PE) as you need

# lvcreate -l number_of_PEs volumegroup -n logvol

This creates a device named /dev/volumegroup/logvol You can then format this device

like any regular disk partition, and then mount the filesystem of your choice on yournew logical volume

RAID Configuration 259

CERTIFICATION OBJECTIVE 5.02

Using Kickstart to Automate Installation

Kickstart is Red Hat’s solution for an automated installation of Red Hat All of thequestions asked during setup can be automatically supplied with one text file Youcan easily set up nearly identical systems very quickly Kickstart files are very usefulfor quick deployment and distribution of Linux systems

There are three methods for creating the required kickstart configuration file:

■ Copy and edit the sample.ks file from the RH-DOCS directory of the Red HatDocumentation CD

■ Use the anaconda-ks.cfg file from the root user’s home directory, /root

■ Use the graphical Kickstart Configurator, accessible through the/usr/sbin/ksconfig command

The first option requires you to copy and modify the kickstart sample file fromthe CD The second option lets you use the kickstart template file created for yourcomputer by Anaconda, the Red Hat Linux installation program The final option,the Kickstart Configurator, is discussed in detail later in this chapter

Copy this kickstart template file and make the appropriate changes suitable foreach generic host (same disk partition scheme, same size of disk or bigger, same videocard and monitor, same mouse, etc.) Place this file where the Anaconda can read it;either locally from the boot disk or remotely from the DHCP/BOOTP server

If you want to put the file on a DHCP/BOOTP server, open the /etc/dhcpd.confconfiguration file on that server Specify the kickstart file For example, if the kickstart

file is on the DHCP server in the /usr/kickstart directory, add the filename

“/usr/kickstart/” command to dhcpd.conf, where filename is the name of the kickstart

file that you want to use Then the setup program will look in the DHCP server’s

directory for a client_ip-kickstart file Alternatively, if the kickstart file isn’t stored

on the DHCP server, add the next-server servername option to the dhcpd.conf

configuration file

No matter where you choose to put the kickstart file, you will typically boot with

a floppy You can even put the kickstart file on the Red Hat boot floppy Just mount

the floppy and copy it as the ks.cfg file from wherever you’ve saved your kickstart

configuration file

At the time of this writing, boot disks still bring up the LILO prompt Use LILO’sboot options to initiate the kickstart file To boot, and then perform the kickstartinstallation from a floppy, type

boot: linux ks=floppy

Installing Red Hat Linux using a kickstart file on a boot floppy is perhaps the easiest method As long as you have a DHCP server on your network, you may

be able to use the same Red Hat Linux installation boot floppy Just boot each new Linux computer from that floppy, type in thelinux ks=floppycommand

at the boot prompt, give the computer a moment to read the ks.cfg file, and insert the same floppy in the next new Linux computer.

If you’re booting from the Red Hat installation CD-ROM, you can still refer to akickstart configuration file on a floppy or hard disk with the following commands:

boot: linux ks=hd:fd0/ks.cfg boot: linux ks=hd:hda2/home/mj/ks.cfg

This assumes the kickstart configuration file is called ks.cfg and is located on the firstfloppy or the second partition of the first IDE drive in the /home/mj directory

You don’t need to get a kickstart file from a DHCP server To boot from a specificNFS or HTTP server on the network, say with an IP address of 192.168.17.18, fromthe /kicks/ks.cfg file, type one of the following commands:

boot: linux ks=nfs:192.168.17.18:/kicks/ks.cfg boot: linux ks=http:192.168.17.18:/kicks/ks.cfg

Most of the options in the sample kickstart file are self-explanatory Every option

is in the sample file and is well commented Follow these ground rules and guidelines

to use when setting up a kickstart file:

■ Do not change the order of the options.

■ You do not need to use all the options

■ If you leave out a required option, the user will be prompted for the answer

■ For upgrades, you must have the following options defined:

■ Language

■ Installation method

Using Kickstart to Automate Installation 261

■ Device specification

■ Keyboard setup

■ The upgrade keyword

■ Bootloader (GRUB or LILO) configuration

If you leave out an option, you will be prompted to complete it This is an easy way to see if your kickstart is configured correctly But as some kickstart options change the partitions on your hard drive, even testing this file can delete all

of the data on your computer So make sure to have a test computer available

to test your kickstart configuration file.

To understand Kickstart, it may be helpful to look through the sample kickstart filefrom the Red Hat Documentation CD, as shown in Figures 5-1 through 5-4 While

FIGURE 5-1

Sample Kickstart

Installation File,

Part 1

this sample file is pretty self-explanatory, a couple of key options need furtherexplanation.

Network

Most options are obvious, but static network configuration options (which requiremanually configuring the IP settings) need special attention Here is an example of

a static network configuration:

network bootproto static ip 172.16.16.5 netmask 255.255.255.0

gateway 172.16.15.254 nameserver 172.16.16.1

Please note that all options must be on one line Line wrapping, if the options exceed

the space in your editor, is acceptable

Using Kickstart to Automate Installation 263

FIGURE 5-2

Sample Kickstart

Installation File,

Part 2

If the client has SCSI, Ethernet, or non-ATAPI CD-ROM devices, don’t forget tospecify them For example:

device scsi aha154x

If you have modules that require a specific IRQ port and I/O address, you can alsospecify those For example:

device ethernet 3c509 opts "io=0x330, irq=7"

Add the continue option if you have more than one device, such as two differentSCSI adapters or Ethernet cards

You have full control of the partitioning options, too You can clear all partitions

with clearpart all, or just clear any Linux-type partitions clearpart linux, or just

add to the end of the current partitions You can create partitions on more than onedrive, but you need to identify each device specifically

FIGURE 5-3

Sample Kickstart

Installation File,

Part 3

Using Kickstart to Automate Installation 265

To add Linux partitions, use the “part” command with the following syntax:

part mount dir size size [ grow] [ maxsize size]

The size is in megabytes You can use the grow option to allow the partition to

expand and fill all remaining disk space (or share it with any other partitions marked

“grow” on the same disk) This will not expand on the fly; but rather, when all size partitions are added, these “growable” partitions will use the rest of the space

fixed-If you specify multiple partitions with the grow option, their space will be dividedevenly You can also specify a maxsize, which will allow the partition to grow only

to the size specified in megabytes

There is one more important issue in a kickstart file: when you specify the rootpassword, encrypt it as follows:

rootpw iscrypted Your_encrypted_password

FIGURE 5-4

Sample Kickstart

Installation File,

Part 4

Then copy the actual encrypted root password from /etc/shadow or /etc/passwd andpaste it into this file.

Kickstart is an essential concept to understand, since installation is such an importantpart of the RHCE exam

EXERCISE 5-2

Creating a Sample Kickstart File from a Running System

for a Second Similar System Installation

In this exercise, you will use the anaconda-ks.cfg file to duplicate the installation fromone computer to another with identical hardware This exercise simply installs all theexact same packages with the same partition configuration on the second computer.Assume that both computers use DHCP to set up their IP addresses Add one rootpartition of 1300MB with a 64MB swap file (This means you must have at least1364MB of space available If you have more space, let it remain unused for now.)You want to install all the same packages as your current installation, so you do notneed to make any package changes to the default anaconda kickstart file in the /rootdirectory

If you do not have multiple computers for this exercise, one alternative as discussedearlier is VMWare

1 Review the /root/anaconda-ks.cfg file Copy it to ks.cfg

2 Create the boot disk and add this ks.cfg file to it:

dd if=/mnt/cdrom/images/bootnet.img of=/dev/fd0 mount /dev/fd0 /mnt/floppy

cp ks.cfg /mnt/floppy

3 Prepare the second computer so that it has the same disk configuration,the same C: drive size if it was present, and same amount of unused andunpartitioned space as the first computer Reboot the second computer withthe Linux boot file with the kickstart file in the floppy drive and the first binary

CD in the CD-ROM

4 At the Red Hat Installation menu boot prompt, enter the following startupcommand:

boot: linux ks=floppy

You should now see the system installation creating the same setup as the first system.You may be asked to put the second binary CD into the CD-ROM drive, if required

Using Kickstart to Automate Installation 267

OPTIONAL EXERCISE 5-2

Modify the Packages to Be Installed

Edit the ks.cfg file on the floppy and remove all the game packages you can find.Alternatively, set up an interactive installation so that you can test and observe theresult on another computer

The Kickstart Configurator

Now that you understand the basics of what goes into a kickstart file, it’s time to solidifyyour understanding through the graphical Kickstart Configurator Go into the LinuxGUI Open a command line interface shell Start the Kickstart Configurator with the

/usr/sbin/ksconfig command The following sections should look familiar, as they

are based on the choices you made when you installed Red Hat Linux

The first section is Basic Configuration, as shown in Figure 5-5 As you can see,this allows you to set up the Language, Keyboard, Mouse, Time Zone, and the language

of the kernel modules associated with this installation By default, the system is to berebooted after Linux is installed, which should lead to a Linux login screen Installation

in text mode installation may be faster, especially if the subject computers have limitedgraphics capabilities Interactive mode is appropriate if you need to edit the ks.cfg filebefore applying it to several other computers And remember to enter a root password.Installation cannot proceed automatically without it

The next section lists Boot Loader Options, as shown in Figure 5-6 The two bootloaders available to Linux are GRUB and LILO Linux boot loaders are normallyinstalled on the MBR If you’re dual-booting with Windows NT/2000/XP with LILO,you need to set up the Windows boot loader to point to LILO on the first sector ofthe Linux partition with the files in the /boot directory Kernel parameters allow you

to pass commands to LILO as described in Chapter 3 If you’re using LILO, linearmode is also known as Large Disk mode in many BIOS menus, and lba32 mode oftenallows Linux to look beyond the 1024th cylinder for Linux /boot files And if you’reupgrading Red Hat Linux, the Upgrade boot loader option allows you to upgrade thepackage associated with your currently installed boot loader

The Installation Method options shown in Figure 5-7 are more straightforward thanwhat is shown in the Red Hat installation program You’re either installing Linux forthe first time or upgrading a previous installation Since you’ll be selecting your ownpackages, the preselected package sets associated with Workstation, Server, or Laptopinstallations are not relevant The installation method is based on the location of theinstallation files.

The Partition Information options shown in Figure 5-8 determine how thisinstallation configures the hard disks on the affected computers The Clear MasterBoot Record option allows you to wipe the MBR from an older hard disk that might

have a problem there; it sets up the zerombr yes command in the kickstart file.

Using Kickstart to Automate Installation 269

Don’t use the zerombr yes option if you want to keep an alternate such as the

files to be a primary partition If you have multiple hard drives or partitions, you maywant to specify the drive or partition for the Mount Point And unless you’re upgrading,you should format each partition.

The Network Configuration section shown in Figure 5-10 enables you to set upthe IP configuration for this specific computer Since kickstart files are generally used

to install Linux on multiple computers, you don’t want to have to specify different IPaddress information for each computer DHCP servers can serve this purpose Supportfor different network configuration options is not complete; if the DHCP server is on

a remote network, you’ll have to go into the ks.cfg file afterward and change “dhcp”

in thenetwork bootproto dhcpline tobootp.

The Authentication section, as shown in Figure 5-11, lets you set up two forms

of security for user passwords: Shadow Passwords, which encrypts user passwords inthe /etc/shadow file, and MD5 encryption This section also allows you to set upauthentication information for various protocols:

■ NIS Network Information Service for one login database on a network withUnix and Linux computers on a network

■ LDAP The Lightweight Directory Assistance Protocol is used for certaintypes of databases such as directories

Using Kickstart to Automate Installation 271