Linux Guide to Linux Certification, Second Edition

1
Linux Guide to Linux Certification, Second
Edition

• Chapter 13
• Troubleshooting and Performance

2
Objectives

• Describe and outline common troubleshooting
  procedures
• Identify good troubleshooting practices
• Effectively troubleshoot common hardware-related
  problems
• Effectively troubleshoot common software-related
  problems

3
Objectives (continued)

• Monitor system performance using command-line and
  graphical utilities
• Identify and fix common performance problems
• Understand the purpose and usage of kernel
  modules
• Recompile and patch the Linux kernel

4
Troubleshooting Methodology
Figure 13-1 The maintenance cycle
5
Troubleshooting Methodology (continued)

• Monitoring Observing system areas for problems
  or irregularities
• Proactive maintenance Minimizing chance of
  future problems
• e.g., perform regular system backups

6
Troubleshooting Methodology (continued)

• Reactive maintenance Correcting problems when
  they arise
• Documenting solutions
• Developing better proactive maintenance methods
• Documentation System information stored in a log
  book for future references
• Troubleshooting procedures Tasks performed when
  solving system problems

7
Troubleshooting Methodology (continued)
Figure 13-2 Common troubleshooting procedures
8
Troubleshooting Methodology (continued)

• Two troubleshooting golden rules
• Prioritize problems according to severity
• Spend reasonable amount of time on each problem
  given its priority
• Try to solve root of problem
• Avoid missing underlying cause
• Justify why a certain solution is successful

9
Resolving Common System Problems

• Two categories of problems
• Hardware-related
• Software-related

10
Hardware-Related Problems

• Often involve improper hardware or software
  configuration
• SCSI termination
• Video card and monitor configuration
• POST test alerts
• Loose hardware connections
• IRQ or I/O address conflicts
• View output of dmesg command

11
Hardware-Related Problems (continued)

• Absence of device drivers prevent OS from using
  associated devices
• Kudzu program Detect and install support for new
  hardware
• If hardware device not detected, device driver
  must be configured manually
• HDDs most common device to fail
• Good idea to use RAID

12
Hardware-Related Problems (continued)
Figure 13-3 The kudzu welcome screen
13
Hardware-Related Problems (continued)
Figure 13-4 Configuring new hardware using kudzu
14
Hardware-Related Problems (continued)

• If HDD containing partitions mounted on
  noncritical directories fails
• Power down computer and replace failed HDD
• Boot Linux system
• Use fdisk to create partitions on replaced HDD
• Use mkfs to create filesystems
• Restore original data
• Ensure /etc/fstab has appropriate entries to
  mount filesystems

15
Hardware-Related Problems (continued)

• If HDD containing / filesystem fails
• Power down computer and replace failed HDD
• Reinstall Linux on new HDD
• Restore original configuration and data files

16
Software-Related ProblemsApplication-Related
Problems

• Missing program libraries/files, process
  restrictions, or conflicting applications
• Dependencies Prerequisite shared libraries or
  packages required for program execution
• Programs usually check at installation
• Package files may be removed accidentally

17
Software-Related ProblemsApplication-Related
Problems (continued)

• rpm V command Identify missing files in a
  package or package dependency
• ldd command Display shared libraries used by a
  program
• ldconfig command Updates /etc/ld.so.conf and
  /etc/ld.so.cache files

18
Software-Related ProblemsApplication-Related
Problems (continued)

• /etc/ld.so.conf file List of directories
  containing shared libraries
• /etc/ld.so.cache file Contains location of
  shared library files
• compressor/decompressor (codec) file Contains
  rules to compress or decompress multimedia
  information

19
Software-Related ProblemsApplication-Related
Problems (continued)

• Filehandles Connections that programs make to
  files
• ulimit command Modify process limit parameters
  in current shell
• Can also modify max number of filehandles
• /var/log directory Contains most system log
  files
• If applications stop functioning due to
  difficulty gaining resources, restart using SIGHUP

20
Software-Related ProblemsOperating
System-Related Problems

• Most software-related problems related to OS
• Boot loader, filesystem, serial device problems
• LILO problems Place linear in, remove
  compact from /etc/lilo.conf file
• GRUB problems Typically result of missing files
  in /boot directory
• mkbootdisk command Create a boot floppy diskette

21
Software-Related ProblemsOS-Related Problems
(continued)

• If filesystem on partition mounted to noncritical
  directory becomes corrupted
• Unmount filesystem
• Run fsck command with f (full) option
• If fsck command cannot repair filesystem, use
  mkfs command to re-create the filesystem
• Restore filesystems original data

22
Software-Related ProblemsOS-Related Problems
(continued)

• If / filesystem is corrupted
• Boot from first Red Hat Fedora installation CD
• Type linux rescue at welcome screen
• Enter shell for Linux system on CD
• Create new / filesystem via mkfs command
• Restore original data to re-created / filesystem
• Reboot system

23
Software-Related ProblemsOS-Related Problems
(continued)
Figure 13-5 The Red Hat Fedora Linux
installation welcome screen
24
Software-Related ProblemsOS-Related Problems
(continued)
Figure 13-6 Obtaining a shell in rescue mode
25
Software-Related ProblemsOS-Related Problems
(continued)
Figure 13-7 The command-line shell used in
rescue mode
26
Software-Related ProblemsOS-Related Problems
(continued)

• Knoppix Linux and BBC Linux Bootable CD-based
  Linux distributions containing many filesystem
  repair utilities
• setserial command Set IRQ, I/O address, and
  speed of serial devices

27
Software-Related ProblemsOS-Related Problems
(continued)
Table 13-1 Common keywords used with the
setserial utility
28
Performance Monitoring

• Jabbering Failing hardware components send large
  amounts of information to CPU
• Other causes of poor performance
• Software monopolizes system resources
• Too many processes
• Too many read/write requests to HDD
• Rogue processes

29
Performance Monitoring (continued)

• Bus mastering Peripheral components perform
  tasks normally executed by CPU
• To increase performance
• Add RAM
• Upgrade to faster HDDs
• Disk Striping RAID
• Decrease kernel size

30
Performance Monitoring (continued)

• Run performance utilities on a regular basis
• Record results in a system log book
• Eases identification of performance problems
• Baseline Measure of normal system activity

31
Monitoring Performance with sysstat Utilities

• System Statistics (sysstat) package Common
  performance monitoring utilities
• Multiple Processor Statistics (mpstat) utility
  Displays CPU statistics
• Input/Output Statistics (iostat) command
  Displays block device input/output statistics
• System Activity Reporter (sar) command Displays
  various system statistics

32
Monitoring Performance with sysstat Utilities
(continued)
Table 13-2 Common options to the sar command
33
Monitoring Performance with sysstat Utilities
(continued)

• Large number of pages sent to and taken from swap
  partition
• Slower performance
• Add RAM to resolve

34
Other Performance Monitoring Utilities

• free command Displays memory and swap statistics
• vmstat command Displays memory, CPU, and swap
  statistics

35
Customizing the Kernel

• Options to provide additional hardware support or
  change existing hardware support
• Insert modules into kernel
• Recompile kernel
• Download and compile new kernel

36
Kernel Modules

• May insert device drivers and kernel features
  into the kernel as modules
• Reduces kernel size
• Good form to compile standard device support into
  the kernel
• Leave support for other devices and features as
  modules
• Modules typically stored in subdirectories of
  /lib/modules/ltkernel-versiongt

37
Kernel Modules (continued)

• insmod command Insert modules into kernel
• modprobe command Insert module and all necessary
  prerequisite modules into kernel
• lsmod command Lists modules currently used by
  kernel
• Can also show module dependencies
• rmmod command Remove modules from kernel

38
Kernel Modules (continued)

• Modules usually inserted into kernel
  automatically at boot time using modprobe
• /etc/modprobe.conf file Used to load any alias
  modules at system startup

39
Compiling a New Linux Kernel

• Gain or remove hardware or kernel support
• /usr/src/ltkernel-versiongt directory Contains
  Kernel source code for a specific distribution
• Can download new kernel source code
• /usr/src/linux directory Contains Kernel source
  code
• Symbolic link to /usr/src/ltkernel-versiongt

40
Compiling a New Linux Kernel (continued)

• Make commands Compilation-related tasks
• Many types
• make mrproper Remove files created by previous
  kernel
• make oldconfig Record current kernel features
  and settings
• make config Prompts user for kernel
  configuration information
• Text-based

41
Compiling a New Linux Kernel (continued)

• make menuconfig Provides menus to select kernel
  configuration
• make xconfig or make gconfig Provide graphical
  interface to select kernel configuration
• xconfig runs in KDE gconfig runs in GNOME
• make clean Remove files not required for
  compilation

42
Compiling a New Linux Kernel (continued)
Figure 13-8 The make menuconfig interface
43
Compiling a New Linux Kernel (continued)
Figure 13-9 The make gconfig interface
44
Compiling a New Linux Kernel (continued)
Figure 13-10 Configuring power options in the
Linux kernel
45
Compiling a New Linux Kernel (continued)

• make bzImage Compile the kernel
• Creates bzip2-compressed kernel
• Copy to /boot directory
• Rename as vmlinuz-ltkernel versiongt
• makemodules_install Compile necessary modules
  and copy to appropriate location

46
Patching the Linux Kernel

• If not changing kernel version, can apply patches
• Rather than download all kernel source code
• Patch command Supply a patch to kernel source
  code
• Still need to recompile kernel

47
Summary

• Administrators monitor the system, perform
  proactive/reactive maintenance, and document
  system information
• Common troubleshooting procedures involve
  isolating and determining the cause of system
  problems and implementing and testing solutions
  that can be documented for future use
• System problems can be categorized as hardware-
  or software-related
• IRQ conflicts, invalid hardware settings, absence
  of kernel support, and hard disk failure are
  common hardware-related problems

48
Summary (continued)

• Software-related problems can be
  application-related or OS-related
• Absence of program dependencies or shared
  libraries, program limits, and resource conflicts
  are common application-related problems, whereas
  boot failure, filesystem corruption, and the
  misconfiguration of serial devices are common
  OS-related problems
• System performance is affected by a variety of
  hardware and software factors

49
Summary (continued)

• The sysstat package contains many useful
  performance monitoring commands
• System features and hardware support are compiled
  into the kernel or provided by modules
• You can compile a Linux kernel with only the
  necessary features and support to increase system
  performance