After the exam I would say that 2 weeks is enough if you have knowledge of common configuration of linux computers. I mean not only the simple installation of nowadays distributions but the manual installation steps beginning with creating partitions until configuration the X files.

Chapter 1.101.1 Basic Bios properties (exam weight: 1)


As important file is here named /proc/pci but this file doesn't exist in my Ubuntu distribution.

IRQ 0 Timer
IRQ 1 Keyboard
IRQ 2 IRQ-Cascading to communicate with IRQ 8-15
IRQ 3 /dev/ttyS1 alias com2
IRQ 4 /dev/ttyS0 alias COM1
IRQ 5 /dev/lp1 alias LPT2
IRQ 6 Floppy controller
IRQ 7 /dev/lp0 alias LPT1
IRQ 8 RTC (real time clock)
IRQ 9 free
IRQ 10 free
IRQ 11 free
IRQ 12 free or PS/2 mouse
IRQ 13 free
IRQ 14 first IDE controller
IRQ 15 second IDE controller

Chapter 1.101.5 Installation of PC cards (weight: 3)

Important things:

Information about current assignment: /proc/interrupts and /proc/ioports and "lspci -vv"


8-Bit-scsi has three address lines. So we have 2^3=8 IDs. The controller gets ID 7 and devices 0-6. Make clear if you being asked for number of ids or devices.

16-Bit-scsi has four address lines. Controller gets also ID 7.

Information files: /proc/scsi/scsi Programs: scsi_info

Device names:

scsi cd or dvd device
scsi raw access. That is for burning devices.
scsi harddisks, usb and sata devices. N means the partition on this device.
scsi generic is the deprecated predecessor of /dev/srX
scsi tape device
Also for tape device but without rewind operation at the end of the job (n=norewind)
Plug and Play

with the program pnpdump we get the configuration and then we write it down into /etc/isapnp.conf i.e. the i/o addresses and interrupts that the pnp card uses. At last we use this properties with program isapnp to activate pnp card.

 $ sudo hdparm /dev/hda

 multcount    = 16 (on)
 IO_support   =  0 (default 16-bit)
 unmaskirq    =  0 (off)
 using_dma    =  1 (on)
 keepsettings =  0 (off)
 readonly     =  0 (off)
 readahead    = 256 (on)
 geometry     = 65535/16/63, sectors = 117210240, start = 0

 hdparm -c1 /dev/hda
activates 32-bit i/o support
 hdparm -d1 /dev/hda
activates dma access for hard disk 1
 hdparm -d0 /dev/hda
disables dma access for hard disk 1
 hdparm -I /dev/hda


 hdparm -i /dev/hda
returns tecnical hard disk data, geometry, serial and model number. With (-I) directly from hard disk bios.

HCF Modems are special modem devices for windows os. Therefore they use specific drivers and so they are often not useable under linux.



you can set the speed of the communication with the serial interface. Wrong speed properties lead to wrong communication and many lost characters.


To connect usb devices to your computer there exists "hotplug" which is also used for pcmcia, firewire devices. Some distributions use "usbmgr".

The kernel modules are called "usb-uhci (Universal Host Controller Interface) for intel and via chipsets and "usb-ohci" (Open Host Controller Interface) for all the other chipsets.

Chapter 1.102.1-6

common question from this part:

Chapter 1.102.1 Creation of a partition set for linux


These partitions have to be placed into the root partition. The installation needs at least two partitions: "/" and "swap" partition. Own Partitions are recommended for

 /boot # as first partition if boot manager is lilo to solve 1024 cylinder problem
FHS (Filesystem Hierarchy Standard)
  1. static files e.g. /usr (for files which seldom are changed)
  2. variable files /var (for files which are often changed)
  3. shareable files (not host specific)
  4. unshareable files (host specific)

Combination of one of the first pair with one from the last pair is also possible. "/home" can be declared as "variable shareable". For this reason we have execution files in /bin as well as in /usr/bin. We have to prevent some files like "mount" in /bin or a mount of /usr partition cannot be made because the binary lies in the partition that we want to mount.

Chapter 1.102.2 Installation of a boot manager

LILO stores the position of kernel and necessary files by means of sector addressing of the hard disk. Therefore lilo has to be reinitialized each time any file has changed a little bit. It is unindependently from naming because lilo does not use inodes to find the file like grub does. Grub uses the common file systems like ext2/3 or reiserfs to get the files. Therefore it is not necessary to reinstall grub into mbr after changing any file in /boot directory.

LILO Errorcodes:

  1. L<errorcode>
first stage was loaded and executed but start of the second stage failed
  1. LI
second stage was loaded but cannot be executed. Maybe /boot/boot.b was moved inside the file system.
  1. LIL
second stage was executed but the map file cannot be found
  1. LIL?
second stage was loaded but the address derives to a invalid point. Maybe because /boot/boot.b was moved.
  1. LIL-
The description table cannot be read.
  1. LILO
all stages were executed successful.

After the bios starts the code in the mbr. Lilo as well as Grub have only a small part in the mbr part. With this small part they load the second part from /boot which loads the kernel into memory.

installs the first stage loader into the mbr and it loads the rest of grub from /boot/grub. That is grub.conf or menu.conf.
INSTALLATION without grub-install with grub console:
 install (hd1,1)/boot/grub/stage1 d (hd2) (hd1,1)/boot/grub/stage2 p (hd1) (hd1,1)/boot/grub/menu.lst

or with

 grub>root (hd1,1)
 grub>setup (hd1)
This is gonna tell you whether or not the stage files can be found and further information about installation process.
installs lilo data into mbr. That has to be done ALWAYS after changes to /etc/lilo.conf

Important files:

Chapter 1.102.3 Installtion from source package

 tar -xzf source.tar.gz


 gunzip source.tar.gz; tar xf source.tar

 make install


 ./configure && make && make install

 make clean

 gzip -d tux.gz
 gunzip -d tux.gz
make both sense because gunzip is often the same file or connected with a symbolic link. The difference recognizes gzip through command line call. "-d" means decompress and is therefore all what gzip needs to know.

With gzip you can also compress a stream of data but not pack files into a package. For this cases we have "tar".

Options for "tar":

create file
extract file
use file which is given as parameter
decompress gzip file
decompress bzip2 file
test mode, what means that the file will be tested for errors but returns also a full list of all files in the packages.

To decrompress file with means of pipes use:

 $bunzip2 src.tar.bz2 | tar -xv -
"-" shows tar that STDIN should serve as input file.

Default parameters for "./configure" script is for example "--with-prefix=/usr/local" which means that the installtion path is "/usr/local".

Default rules for make files:

deletes all .o files and other temporary files
clean all the data so that we have the origin status as we have right after the tar command. The rule for the source files of the kernel is called "mrproper".
deprecated, because we have often the "configure" script.
handles dependencies
starts the compile process.
starts the installation process.
tries to bring up an already installed version up to date

Chapter 1.102.4 Management of shared libraries

shared objects: *.so

 ldd /bin/echo
 linux-gate.so.1 =>  (0xffffe000)
 libc.so.6 => /lib/tls/i686/cmov/libc.so.6 (0xb7e19000)
 /lib/ld-linux.so.2 (0xb7f5e000)

File /etc/ld.so.conf define where the system search for libraries. Some distributions use also files from /etc/ld.so.conf.d/. No prevent searching whole list of pathes it exists an cache. With "ldconfig" you can bring this cache up to date.


Additionally the variable LD_LIBRARY_PATH gives pathes which will be searched for necessary libraries.


Chapter 1.102.5 Debian package administration

 dpkg -i
 dpkg -r or dpkg --remove
 dpkg -l
list of installed packages
 dpkg -l 'gnome*'
list all installed packages that begins with "gnome"
 dpkg -L package
shows all files from package
 dpkg -S pattern or dpkg --search pattern
search for pattern in list of installed files.
 dpkg -P package 
or "--purge" means that all remaining config files will also be deleted.
 dpkg -s package or dpkg --status package

 dpkg -E
avoids installation of packages of the same version as the one that is already installed
 dpkg -G
avoids installation of an older package over one that is newer.
 dpkg -R
traverses recursively all sub folders and installs all packages .deb that will be found.
 apt-get update
load recent package list
 apt-get upgrade
install new versions if any are available.
Menu based install tool.
converts .rpm to .deb an vice versa.

Config files:

  • /etc/dpkg/dpkg.cfg
  • /etc/apt/*

Important files:

Chapter 1.102.6 RPM

 -i --install
install package
 -e --erase
erase package
 -U --upgrade
upgrade package
 rpm -i package.rpm --test
test mode
ignores error respectively wrong dependencies
for installation of updates. Option --oldpackage approves installation of old version above newer ones.
 -V --verify
test package and shows lacking and modified files.
 -K --checksig
Run some tests before installation as checksum or signature test.
 -q --query
Show extra information.
List of all files which are installed.
List of all config files which are connected to the package.
Manual pages
Common information about the package.
List of all packages which are installed.
 -qf file
returns package name which contains the given file
 $rpm -qf /usr/bin/passwd

If the package will be installed later you can add option "p" to the commands above. "-p packagename" means that the search will not go through installed packages but through files of packagename.

 -c --configfiles
searchs only config files
 -d --docfiles
searchs only documentation files
 -l package --liste package
all files which the package contains.
 -R --requires
searchs only dependencies
 -i (with q as -qi, not -i as --install)
searchs in the information about the package
repairs rpm database

Important files:

Chapter 1.103

Configuration of bash

If you start with a login shell bash searchs for the following files in the given order:

  1. /etc/profiles
  2. ~/.bash_profile
  3. ~/.profile

The file ~/.bashrc is only for shells which were interactive meaning that they were start during execution of another shell.

Chapter 1.103.1 Working on the command line (exam weight:5)

Important things:

Here is basic knowledge of bash or sh necessary like variable declaration, variable exporting, usage of history as !-5 or !234.

env (run a program in a modified environment)
 env [OPTION]... [-] [NAME=VALUE]... [COMMAND [ARG]...]
Set each NAME to VALUE in the environment and run COMMAND.

Chapter 1.103.2 Processing text via filters (exam weight:6)

Important things:

cat (returns the content of the given file)
cut (cut characters or fields from files)
 $cut -option file
 $cut -c5 <file>
extracts from each line the fifth character
 $cut -c5- <file>
extracts from each line the fifth character until the end of the line
 $cut -c5-10 <file>
extracts fifth until tenth character
 $cut -c1,3,5,7 <file>
extracts list of given positions

We can also define a delimiter. That is practical for csv files where the delimiter is the ";" character. The delimiter devide a line into parts or better called as fields. To define the delimiter we have option "-d" and the fields we want to see will be defined with "-f". By default is the delimiter the tabulator \t.

 $cut -d\; -f1,3 file.csv
extracts field 1 and 3
expand (transform tab characters into space characters)

By default it transform one tab to eight space characters. You can choose how many characters one tab should be with option "-" and then the number of spaces. For example to transform tab to four spaces use:

 expand -4 main.c

"-0" is not possible. To delete tabs from file use "tr".

       fmt - simple optimal text formatter


       fmt [-DIGITS] [OPTION]... [FILE]...


       Reformat each paragraph in the FILE(s), writing to standard output.  If
       no FILE or if FILE is -, read standard input.

       Mandatory arguments to long options are  mandatory  for  short  options

       -c, --crown-margin
              preserve indentation of first two lines

       -p, --prefix=STRING
              reformat  only lines beginning with STRING, reattaching the pre‐
              fix to reformatted lines

       -s, --split-only
              split long lines, but do not refill

       -t, --tagged-paragraph
              indentation of first line different from second

       -u, --uniform-spacing
              one space between words, two after sentences

       -w, --width=WIDTH
              maximum line width (default of 75 columns)

       --help display this help and exit

              output version information and exit

       With no FILE, or when FILE is -, read standard input.
head (output the first part of files)
 -n, --lines=[-]N
  print the first N lines instead of the first 10; with the leading -, print all but the last  N  lines  of
  each file
hexdump, hd (ASCII, decimal, hexadecimal, octal dump)
 $hexdump -C a.avi | head 
 00000000  52 49 46 46 90 54 de 01  41 56 49 20 4c 49 53 54  |RIFF.T..AVI LIST|
 00000010  8c 20 00 00 68 64 72 6c  61 76 69 68 38 00 00 00  |. ..hdrlavih8...|
 00000020  57 82 00 00 00 00 00 00  00 00 00 00 10 01 00 00  |W...............|
 00000030  84 03 00 00 00 00 00 00  02 00 00 00 00 00 00 00  |................|
 00000040  80 02 00 00 e0 01 00 00  00 00 00 00 00 00 00 00  |................|
 00000050  00 00 00 00 00 00 00 00  4c 49 53 54 1c 10 00 00  |........LIST....|
 00000060  73 74 72 6c 73 74 72 68  38 00 00 00 76 69 64 73  |strlstrh8...vids|
 00000070  4d 4a 50 47 00 00 00 00  00 00 00 00 00 00 00 00  |MJPG............|
 00000080  e8 03 00 00 12 75 00 00  00 00 00 00 84 03 00 00  |.....u..........|
 00000090  5b c4 00 00 10 27 00 00  00 00 00 00 00 00 00 00  |[....'..........|
join - join lines of two files on a common field
 $cat test1
 testfield1 testfield2
 $ cat test2 
 testfield1 testfield2fromtest2
 $ join test1 test2 
 testfield1 testfield2 testfield2fromtest2
less (print file content pagewise and possibility to navigate through the content)
more (print file content pagewise but you can not navigate through the content)
nl (print file lines with their line number)
 nl -s '.) ' linearerKongruenzGenerator.c | head
     1.) #include <stdio.h>
     2.) #define OBERGRENZE     10
     3.) #define MAX_LOESUNGEN   100

     4.) /*
     5.) unsigned in ggT(unsigned int a, unsigned int b);
     6.) unsigned in euclid(unsigned int a, unsigned int b);
     7.) */

     8.) unsigned int euclid(unsigned int a, unsigned int b)
paste (merge lines from files)
 $cat test1 test2
 testfield1 testfield2
 testfield1 testfield2fromtest2
 $ paste test1 test2
 testfield1 testfield2   testfield1 testfield2fromtest2
pr - convert text files for printing
 $cat test1 test2
 testfield1 testfield2
 testfield1 testfield2fromtest2
converts text content and print it pagewise similar to the following:
 2007-02-27 17:57                      test1                       Page 1
 testfield1 testfield2

 2007-02-27 17:58                      test2                       Page 1
 testfield1 testfield2fromtest2
 sed s/./A/
replaces every first occurence of a arbitrary character with a "A".
 sed s/\\./A/ 
replaces every first occurence of a "." with a "A".
 sed s/\\/x/ (without use of bash)
replaces every first occurence of a "\" with a "x". But the bash replaces "\\" with a "\" because it interprets backslash as an special character. But "sed" needs two backslashes. So we have to give the bash four backslashes. Bash replaces that with "\\" and sed gets its two needed backslashes and can recognize that it should search for a backslash character.
 sed s/\\\\/x/
sort (sort text or files)
 -n (numerical)
 -f (no difference between lower and capital characters)
 -r (reverse order)
 -n -r (reverse numerical order)
 -c (test if file is sorted and returns errorlevel 1 if not)
 -u (no double occurence of lines)
split (split a file into pieces)

Without any parameters "split" creates pieces of 1000 lines.

 $split -100 logfile.log
 xaa xab xac xad 
The logfile was devided into pieces a 100 lines.
 $for file in `ls x* | sort`; do cat $file >> newlogfile.log;done
takes pieces and stick them again together to a new file.

Let us assume that we have a file of size 3 gigabyte. We want to split this into pieces of size 700 megabyte to burn them onto cds.

 $split -b 700m filesize3gb.tgz
After copying files back to harddisk we can concatenate these files together again with:
 $cat filesize3gb* > hugefile.tgz
tac (as cat only with reverse order)

"tac" prints out the last line first then the second last line and so on.

tail (prints end of file)
 $tail -20 messages
shows last 20 lines of file messages
 $tail +100 messages
shows lines from line 100 until end of file.
 $tail +100 -r messages
as tac prints out last line and go further with reverse order.
tee (t-piece for output redirecting)

If you want to redirect stdout to a file and want to see simultaneous stdout on screen "tee" is the right tool.

 $du -bc | sort -n | tee bytesorted.log
writes output to screen and to file bytesorted.log. If you want to print into multiple files just add parameters to "tee".
touch (creates file)
 -a change the access time
 -c if file does not exist it will not be created
 -m change the modified time
tr (replace characters)
 echo "Hello World" | tr [:upper:] [:lower:] 
gives "hello world"
  tr - translate or delete characters
  tr [OPTION]... SET1 [SET2]
  Translate, squeeze, and/or delete characters from standard input, writing to standard output.
  -c, -C, --complement
         first complement SET1
  -d, --delete
         delete characters in SET1, do not translate
  -s, --squeeze-repeats
       replace  each  input  sequence of a repeated character that is listed in SET1 with a single occurrence of
         that character

  -t, --truncate-set1
         first truncate SET1 to length of SET2

  --help display this help and exit

         output version information and exit

  SETs are specified as strings of characters.  Most represent themselves.  Interpreted sequences are:

  \NNN   character with octal value NNN (1 to 3 octal digits)

  \\     backslash

  \a     audible BEL

  \b     backspace

  \f     form feed

  \n     new line

  \r     return

  \t     horizontal tab

  \v     vertical tab

         all characters from CHAR1 to CHAR2 in ascending order

         in SET2, copies of CHAR until length of SET1

         REPEAT copies of CHAR, REPEAT octal if starting with 0

         all letters and digits

         all letters

         all horizontal whitespace

         all control characters

         all digits

         all printable characters, not including space

         all lower case letters

         all printable characters, including space

         all punctuation characters

         all horizontal or vertical whitespace

         all upper case letters

         all hexadecimal digits

         all characters which are equivalent to CHAR

  Translation occurs if -d is not given and both SET1 and SET2 appear.  -t may  be  used  only  when  translating.
  SET2  is extended to length of SET1 by repeating its last character as necessary.  Excess characters of SET2 are
  ignored.  Only [:lower:] and [:upper:] are guaranteed to expand in ascending order; used in SET2 while translat‐
  ing,  they  may only be used in pairs to specify case conversion.  -s uses SET1 if not translating nor deleting;
  else squeezing uses SET2 and occurs after translation or deletion.

Example: Replace linefeed with nothing.

 cat linefeed.txt | tr -d '\012'
uniq (do not print lines that occurs twice)
 $wc test.txt
 671  4168 26487 test1.txt
prints number of lines, words and characters
 $wc -c test.txt
 $wc -w test.txt 
 $wc -l test.txt

"wc" is able to get its input from stdin, pipe or through parameter.

File Globbing (bash replaces * ? and [] with fitting patterns)
whereis (search for files)

"whereis" starts a search for a particular file or command and returns its full path and if possible its manual sites.

xargs (takes a list from stdin and call a command with this list)
 $find . -name "*.tmp" -print0 | xargs -0 rm
searchs for all files with extension .tmp and the list goes to xargs. xargs takes that list and calls rm with the whole list. That means that we have only one call instead of several hundrets if we go through the file list with a loop and calls for every match the "rm" command.
`` (back quotes)
 $ today=`date`
 $ echo $today
 Wed Feb 28 19:58:55 CET 2007
test (to evaluate expressions)
 if test expression
  1. do something

or you can use the symbolic form of "test" with "[ expression ]":

 if [ expression ]
   # do something

The spaces inside the brackets are not a coincidence. After [ and before ] you have to place a space character. and also the expressions will be devided with a space character.

 [ var1 -eq var2 ]
 [ var1 -ne var2 ]
 [ var1 -lt var2 ]
 [ var1 -gt var2 ]
 [ var1 -le var2 ]
 [ var1 -ge var2 ]

 if [ $# -ne 2 ]
   echo "You have to give 2 parameters!"
let (compares integer numbers)

As "test" we have a short form for let:

 (( var1 == var2 ))
 (( var1 != var2 ))
 (( var1 < var2 ))
 (( var1 > var2 ))
 (( var1 >= var2 ))
 (( var1 <= var2 ))
unexpand (converts spaces to tabs)
 Convert  blanks  in  each  FILE to tabs, writing to standard output.  With no FILE, or when FILE is -,
  read standard input.

  Mandatory arguments to long options are mandatory for short options too.

  -a, --all
         convert all blanks, instead of just initial blanks

         convert only leading sequences of blanks (overrides -a)

  -t, --tabs=N
         have tabs N characters apart instead of 8 (enables -a)

  -t, --tabs=LIST
         use comma separated LIST of tab positions (enables -a)

Chapter 1.103.3 File management (exam weight:3)

Important things:

Command cp (copy)
 -d (no dereference)
copyies links as links and not as their reference content.
 -f (force)
If destination files can not be opened for writing this parameter leads to removing the file and its new establishment as it exists in the source
 -i (interactive)
asks each time if a file will be overwritten.
 -l (link)
does not copy the files physically but as hard links. That implies that the destination must be on the same partition.
 -p (preserve attributes)
keeps file attributes as owner, access rights and so on.
 -R -r (recursive)
copyies also subfolders
 -s (symlink)
same as "-l" but not hard links but symlinks.
 -u (update)
overwrites existing destination files only if the source file is newer.
 -v (verbose)
gives more output.
 -Rpd   or -a
copyies recursive, permits rights and does not dereference links. These options are called archiv options and have a short parameter "-a"

Chapter 1.103.4 Using streams, pipes, redirects (exam weight:5)

Important things:

 $cmd1 | cmd2
cmd1 stdout will be connected via pipe with stdin of cmd2
 $cmd1 ; cmd2
first execution of cmd1 then cmd2
 $cmd1 && cmd2
first cmd1 and if it returns errorlevel 0 then execution of cmd2 else nothing happens
 $cmd1 || cmd2
first cmd1 and if it returns errorlevel unequal to zero then execution of cmd2
 $cmd1 & cmd2
cmd1 execution in background then cmd2 in foreground.
 $cmd > test.out or cmd 1>test.out
stdout to test.out
 $cmd >> test.out
stdout appended to test.out
 $cmd 2>error.log
stderr to error.log
 $cmd &>output.log or cmd > output.log 2>&1
both stdout and stderr into output.log
 $cmd < intput.txt
gets stdin from intput.txt
 $cmd << EOT input
cmd gets input from file input until string EOT occurs.
 $cat << EOM | mail user@localhost
 Subject: Hello
bash option "noclobber"
 $set -o noclobber
 $echo Test > test.txt
 bash: test.txt: cannot overwrite existing file
 $set +o clubber
 $echo Test > test.txt

Chapter 1.103.5 Creating, monitoring and killing of processes (exam weight:5)

Important things:

Command "jobs"

shows all processes that are stopped or running in the background.

takes job with id and continue it in the foreground.
 bg <JOBID>
takes job with id and continue it in the background.
Command "ps"
 ps -ef
returns snapshot of current processes with enviroment (e) and full formatted output (f)
 ps aux
returns overview with usage of cpu time and so on.
Command kill

ends the given process.

 kill <pid>
 kill %<jobid>
 kill -9 <pid>
sends signal 9 alias SIGKILL to process.
If a process gets this signal it should end itself.
forces the termination of the process.
used to be as "Hang Up immediately" but nowadays it is used to force process to load its config files again.
 kill -HUP <pid>
forces process pid to load its config files again.
same as we press <ctrl>+z in the console what has the result in a process that is stopped.

Chapter 1.103.6 Modifying process priorities (exam weight:3)

Important things:

 nice -n 18 program
the higher the number the friendlier you are to other users because you give away cpu time. Value interval goes from -20 to 19 because we have 40 levels and 0 is also one level.
 nice -18 <pid>
 nice -+18 <pid>
are the same.
 nice --18 <pid>
sets priority to -18, what is a very high priority.

Chapter 1.103.7 Searching text files via regular expressions (exam weight:3)

Important things:

arbitrary character
begin of a line
end of a line.
0 or one repetition of previous expression. Expression "?" stands for a arbitrary character that occurs once or none.
one or more repetitions of previous expression.
no or arbitrary repetitions of previous expression.

Chapter 1.103.8 vi basics (exam weight:1)

Important things: vi / or ? h,j,k,l G, H, L i, c, d, dd, p, o, a ZZ, :w!, :q!, :e! :!

Chapter 1.104


Chapter 1.104.1 Creation of partitions and file systems

Important things:

Partitions IDs
 $ls -l
 $du -a
to get file size of individual files
 $ls -lr
sorts backwards
 $du -c
gives size of all subfolders and a summary.
 #fdisk -l
 Disk /dev/hda: 60.0 GB, 60011642880 bytes
 255 heads, 63 sectors/track, 7296 cylinders
 Units = cylinders of 16065 * 512 = 8225280 bytes
Device Boot Start End Blocks Id System
 /dev/hda1               1        7166    57560863+  83  Linux
 /dev/hda2            7167        7296     1044225   82  Linux swap / Solaris

To start partition tool fdisk just give as parameter the hard disk device.

 fdisk /dev/hda
gives help a toggle a bootable flag b edit bsd disklabel c toggle the DOS compatibility flag d delete a partition l list known partition types m print this menu n add a new partition o create a new empty DOS partition table p print the partition table q quit without saving changes s create a new empty Sun disklabel t change a partition's system id u change display/entry units v verify the partition table w write table to disk and exit x extra functionality (experts only)
creates new partition. Numbers will be interpreted as absolute cylinder i.e. 150 means that partition ends up with cylinder 150. "+2000M" for size of 2 gb for partition. "fdisk" asks first for the type of the partition i.e. "logical 5 or above" or "primary partiton 1-4". Only four primary partitons can be created. Primary Partitions are defined in the Master Boot Record (MBR). Should it be more than four you have to create an extended partion. In an extended partition you can create further 63 (IDE) or 15 (SCSI) logical partitions. Other operating systems can only be installed into a primary partition. Linux can be installed into both types. Partition types for linux is 83 or 82 (swap).

To define a file system in a partition use mkfs. By default it creates type ext2. Alternatively you can use option -t or use mkfs.reiserfs. For ext3 use mkfs.ext3 or "-t ext3". For windows file system use mkfs.vfat.

 mkfs -t reiserfs
to make a reiser file system.

For swap partitions we have:

If you have forgotten to create space for a swap partition you can create a file as swap "device".

 dd if=/dev/zero of=/var/swap bs=1024k count=128
creates file /var/swap with size 128mb
 chmod 600 /var/swap
 mkswap /var/swap
 swapon /var/swap

To view all swap devices which are being used:

 $ cat /proc/swaps
 Filename                                Type            Size    Used    Priority
 /dev/hda2                               partition       1044216 101804  -1

Chapter 1.104.2 Maintenance of integrity of the file systems (exam weight: 3)

Important things:

From time to time every file system should be tested with fsck or for ext2/3 with program e2fsck.

 -s (serialize)
process file system checks one by one
 -A (all)
tests all file system of /etc/fstab
 -N (no operation/NOOP)
tests but does not change anything.
 -p (preen) or -a (automatic) (latter is deprecated)
tests and repairs automatically.

Chapter 1.104.3 Control of mounting and unmounting of file system

Important things:

 mount /dev/hda5 /mnt/tmp
mounts partition hda5 into current filesystem.
gives overview of mounted partitions
 df -h 
gives a (human readable) overview of mounted partitions with a table of assigned and free space on these partitions.

To unmount partition from file system use

 umount /dev/hda5


 umount /mnt/tmp

To do this automatically create an entry in /etc/fstab. A line as follows mounts partition 5 to /mnt/tmp at the system start.

 /dev/hda5      /mnt/tmp      ext3     defaults     0 0

Another example:

 # <file system> <mount point>   <type>  <options>       <dump>  <pass>
 proc            /proc           proc    defaults        0       0
 /dev/hda1       /               ext3    defaults,errors=remount-ro 0       1
 /dev/hda2       none            swap    sw              0       0
 /dev/hdc        /media/cdrom0   udf,iso9660 user,noauto     0       0

"defaults" means "rw"+"auto" for automatically mount at boot.

The dump column defines for data backup if partition is to be stored or not (1 or 0).

The last column gives the order for a system check after a system crash (2-9).

Swap partitions get mount point "none" and type "swap" as well as options like "sw" and "prio=42" which defines the priority.

The options "usrquota" and "grpquota" are for user quotas.

"devpts", "proc", "usbdevfs" are for internal functions.

For cd-roms we have "noauto" because the device doesn't contain always medias at the start of the system. "user" means that a normal user can mount the partition what make sense for cdrom devices. Option "user" implies that if a specific user mounts a partition only this user can umount the partion again. Option "users" implies that every user can mount and any other user can umount this partition. "exec" is used to are able to execute programs on cd media.

Chapter 1.104.4 Manage disk quotas

Important things:

This starts with entries in "/etc/fstab". In particular the options "usrquota" and "grpquota" are of importance. /etc/fstab:

 /dev/sda6 /var reiserfs usrquota,grpquota  1 2

 mount -o remount /var
to mount partition without reboot
 quotacheck -avugm
checks all partitions with quota options from /etc/fstab
v means verbose
m forces to check a mounted partition
u creates config file aquota.user
g creates file aquota.group in /var/
These files contain information which user has currently how many space allocated.
changes quotas for specific users.
 Filessystem      blocks    soft         hard    inodes    soft    hard
 /dev/sda6          2243     150000  200000     1            0        0
means that user tux has currently 2243 blocks a 1kb allocated. Softlimit is 150mb and hard limit is 200mb. While overstepping the soft limit you have by default seven days remaining to come under this limit. While reaching the hard limit ends every following write access.
activates user quota limits
shows a report of the current quota status.
shows every user his quota status. User root can ask for status of the quota of a specific user with "quota user" .

The limit that reaches first its limit counts. That means that a user cannot write any more if its group limit has been reached although his own limit has not been reached.

There must exist two files in the root level of the partition that you want to mount with quota rules. Once "quota.user" and "quota.group" both with rights 600.

Chapter 1.104.5 Access control for files with attributes

Important things:

first three are for the owner
second three for the group
last three for all others

Sticky bit means for a directory that its files can only be removed from its owner. That makes sense for /tmp. To be recognized as "t" at the position of "x" from others.

For group directories we have SGID (Set Group ID Bit) which appears as "s" instead of "x" at group position. That means that every new file in this directory has the ownership of the group not of the creator. Assigned to a file means that its execution process gets the identity of its group not from the user that execute the file.

SUID (Set User ID Bit) appears as "s" at "x" position of the owner. It brings up the execution process under identity of the owner of the file not from the user that execute the file.

"S" or "T" means that the execution right is missing. That does not make very sense.

 To set these sticky bits use a four digit value with chmod:
 $ chmod 1100 test ; ls -l test
 ---x-----T 1 micki micki 0 2007-02-22 17:29 test
 $ chmod 2100 test ; ls -l test
 ---x--S--- 1 micki micki 0 2007-02-22 17:29 test
 $ chmod 4100 test ; ls -l test
 ---s------ 1 micki micki 0 2007-02-22 17:29 test
 $ chmod 4400 test ; ls -l test
 -r-S------ 1 micki micki 0 2007-02-22 17:29 test

 -r-------- 1 micki micki 134 2006-12-17 16:32 blacklist.txt
 $ chmod o+w blacklist.txt ; ls -l blacklist.txt
 -r------w- 1 micki micki 134 2006-12-17 16:32 blacklist.txt
 $ chmod g+x blacklist.txt ; ls -l blacklist.txt
 -r----x-w- 1 micki micki 134 2006-12-17 16:32 blacklist.txt 
 $ chmod u+x blacklist.txt ; ls -l blacklist.txt
 -r-x--x-w- 1 micki micki 134 2006-12-17 16:32 blacklist.txt 
 $ chmod u-r blacklist.txt ; ls -l blacklist.txt
 ---x--x-w- 1 micki micki 134 2006-12-17 16:32 blacklist.txt
 $ chmod a+r blacklist.txt ; ls -l blacklist.txt
 -r-xr-xrw- 1 micki micki 134 2006-12-17 16:32 blacklist.txt
 chmod 712 blacklist.txt ; ls -l blacklist.txt
 -rwx--x-w- 1 micki micki 134 2006-12-17 16:32 blacklist.txt


 $ umask 0044; touch test;   ls -l test;rm test
 -rw--w--w- 1 micki micki 0 2007-02-22 17:11 test 
 $ umask 0077; touch test;   ls -l test;rm test
 -rw------- 1 micki micki 0 2007-02-22 17:12 test
 $ umask 0022; touch test;   ls -l test;rm test
 -rw-r--r-- 1 micki micki 0 2007-02-22 17:12 test
 $ umask 0000; touch test;   ls -l test;rm test
 -rw-rw-rw- 1 micki micki 0 2007-02-22 17:13 test
The "x" attribute won't be set while creation whatever value umask has for security reasons.

For directories and its "x" flag use the three digit parameter:

 $ umask 000; mkdir test;   ls -ld test;rmdir test
 drwxrwxrwx 2 micki micki 4.0K 2007-02-22 17:19 test
 $ umask 234; mkdir test;   ls -ld test;rmdir test
 dr-xr---wx 2 micki micki 4.0K 2007-02-22 17:20 test
 $ umask 127; mkdir test;   ls -ld test;rmdir test
 drw-r-x--- 2 micki micki 4.0K 2007-02-22 17:20 test

Some special attributes for ext2 file systems for example the immutable bit which can only be set by root. After it was set you cannot delete this file. To set this bit use "chattr":

 chattr +i importantfile
Because "ls -l" does not show this bit you can use "lsattr":
 lsattr importantfile

Chapter 1.104.6 Manage file owner

Important things:

After creation of a file, the owner is the current user. Only root is allowed to change its owner with

 chown user filename

A normal user can only change the group definition of a file and this only if he is owner of the file and member of the new group.

 chgrp users filename.txt

Chapter 1.104.7 Creating and changing of hard respectively symbolic links

Important things:

 ln -s destination source
creates a symbolic link. A symbolic link can be recognized by "l" in front of the access attributes after command "ls -l" and by showing file name "source -> destination"

To create an hard link we use also ln but without parameter. The destination cannot be an directory only a file. The os saves the content of a file in blocks on the hard disk. Inodes are deriving on this blocks. The inodes contain also additional informations like owner, access attributes, timestamp and so on. The hard link is a file that derives on the inode of the destination file. So the link keeps valid if the file is moved inside the same partition. Hard links are only valid inside the same partiton because each partition has its own inode management.

 ln source destination
 ls -li
 6153239 -rw-r--r-- 2 micki micki 66K 2007-02-21 10:14 destination
 6153239 -rw-r--r-- 2 micki micki 66K 2007-02-21 10:14 source
source and destination point both on inode 6153239. If you remove one file the content of the file is not been lost.
 $ ls -li blacklist.txt
 6153522 -rw-r--r-- 1 micki micki 134 2006-12-17 16:32 blacklist.txt
 $ ln blacklist.txt hardlink1
 $ ls -li blacklist.txt
 6153522 -rw-r--r-- 2 micki micki 134 2006-12-17 16:32 blacklist.txt
 $ ln blacklist.txt hardlink2
 $ ls -li blacklist.txt
 6153522 -rw-r--r-- 3 micki micki 134 2006-12-17 16:32 blacklist.txt
As you see the third column shows the number of names a file or inode has. If you delete one the number decrease but the content of the file keeps unchanged until the last reference or name to this inode has been deleted. The rules for directories are a little bit different. The counter is at least 2 because in every directory itself lies the file with name "." which points at the current directory. With every sub directory the counter increase because each one has the file ".." which points at the father directory.
 srwx------ ... amavisd
is the socket from AMaViS.
 chmod u+rwx,g+rw <filename>

and not

 chmod u+rwx g+rw <filename>

 a (append only)
file cannot be deleted. Only content can be appended.
 c (compressed)
not supported
 i (immutable)
File is unchangeable. Cannot be deleted, renamed or changed nor a symlink can be derived to this file.
 s (secure deletion)
If the file will be deleted the sectors of the hard drive will be filled with random numbers instead of only delete its reference in the file system table.
 S (synchronous update)
The file will be flushed and not cached by the hard drive cache.
 u (undeleteable)
Should appear as opposite to "s" but is not supported by the kernel yet.

To set these attributes use:

 chattr +a <filename>
sets attribute "a"
 chattr -a <filename>
unsets "a"

Chapter 1.104.8 Finding and placing files at the right position in file system

Important things:

 $which ping
"which" search for the given file in the search path $PATH.
 $whereis ping
 ping: /bin/ping /usr/share/man/man8/ping.8.gz
gives the path independent from value of PATH and additionally the manual file.
 $find / -type f -name '*.mp3'
searchs for mp3 files from root directory to its sub directories.
 $find -type f -iname '*~' -exec rm {} \;
removes temporary files. "-iname" ignores lower and capital character. "{}" will be replaced with the file name that has been found. To find a specific inode and its file you can use "-inum". Without the "\" before ";" the bash would interpret ";" for itself and find would not see it and so the syntax of the exec parameter would be hurted.
 $find ~ -mtime 2
finds files that were modified 2 days ago. find gets its time parameter ALWAYS in days!
 $find ~ -atime -3
finds files that were accessed in the time period back to 3 days. That is 2 minutes ago, 12 hours ago and also 3*24 hours ago.
 $find ~ -atime +3
finds files that were accessed in the past and at least 3 days ago.
means access time so we have also readings that change this timestamp.
means modification time. That is the time when last writing access has occured to the file.
means change time. That is the time when the last changes were made to the file information as access rights, owner and so on.
 find / -perm +6000
finds all files that have at least ("+") the rights 6000.

If file names contains spaces we have a problem if we put them through pipe to "xargs". A solution is to define the end of the file as zero character.

 find / parameter -print0 | xargs -0 command

"xargs" has a parameter "-i" to define a pattern that stands for the name "xargs" gets from stdin. That is an advantage if you need the pattern twice for example if you rename the file.

 xargs -ixxx mv xxx /tmp/rename-xxx < filelist.txt
reads the file list from file filelist.txt and substitutes the pattern xxx with the current name from filelist.txt. Between option an pattern must not stand a space character!
 $locate pdf
searchs in a internal database for file names and its pathes for the given pattern. Possible results are a.pdf or /pdfs/a.doc. The latter one because "pdf" is also part of its path "/pdfs".

Use "updatedb" to update this database.

 ls -l
shows also the type of the file in the first column. The character have the following meaning:
 d directory
 l symlink
 c character device
 b block device
 p fifo
 s socket

Chapter 1.110 X

Chapter 1.110.1 DISPLAY MANAGER: Installation and Configuration of X11 (exam weight:5)

Important things:

 Xorg -configure # automatic configuration 
 xorgcfg         # graphical tool
 xorgconfig      # console tool

Results will be written to /etc/X11/xorg.conf. You activate the X server with "startx"

If the screen is a little bit hidden or if you want to change size of the viewed screen use "xvidtune". After you are happy with "xvidtune" this program returns a modeline like:

 Modline "1280x1024" 166.29 1280 1392 1528 1744 1024 1025 1028 1077
Col1: Name of the mode
Col2: pixel clock (factor between pixel and seconds)
Col3-6: values for horizontal timing
Col7-10: values for vertical timing

The "xorg.conf" file is devided into sections. Each section begins with "Section" and ends with "End Section". The different types of section are the following ones:

The X font server "xfs" serves the font wishes of the X server. The config file is "/etc/X11/xfs.conf" or "/usr/X11R6/lib/X11/fs/config". With an entry as 'FontPath "tcp/"' in the Files section the X server tries to get the fonts from the specified server.

After installing new fonts you have to call "mkfontdir <path>" and "mkfontscal2 <path>" to be able using the fonts.

ATTENTION: it does not exist a section "Font"!

Chapter 1.110.2 Installation of the display manager (exam weight:3)

Important things:

The display manager presents a login dialog and after user login it starts a window manager like KWin (KDE), Metacity (Gnome). The well known display manager are XDM, KDM, GDM.

Configuration of the greetings message of login screen lies in "/etc/X11R6/*/Xresources". Inside this files we have values as "Client*Resource: value". For example to start xterm with a given geometry use "XTerm*geometry:90x40". Then xterm will be started each time with "xterm -geometry 90x40".

The protocol "XDMCP" (X Display Manager Control Protocol) is for remote access over network. After that you can take an old rusty pc and call the X with "X :0.0 -query <host>" where host is the one that should serve the data. After that you are working on the old pc with X which comes from the host over network. The old pc takes only the input from keyboard and mice and sends it to the host. For that reason the remote host must be a strong processing unit, if it has to process different user over network.

 /usr/X11R6/bin/gdm (binary)
 /usr/X11R6/lib/X11/xdm (config file)

To allow remote connections you have to uncomment the following line in xdm config file with a "!":

 DisplayManager.requestPort: 0
 !DisplayManager.requestPort: 0

Config files under "/etc/X11/gdm".


To adjust these files you can use "gdmsetup".

 kdmrc (config file under KDE_DIR/share/config/kdm)

Chapter 1.110.4 Installation and adjustment of a WINDOW MANAGER (exam weight:5)

Important things:

Window Manager:

To adjust these use file "~/.Xdefaults". The syntax is as follows:

 program.class.resource: value

"class" means for example "Geometry" for common properties for window size and window position.

 xterm*Background: black
xterm windows get a black background.
 xhost +
allows host to open windows on the local x server. To derive the output from to the local host start the application with parameter "-display" or set enviroment variable "DISPLAY" as "Host:Xserver.Display"
 xeyes -display tux:0.0
start xeyes on screen of host tux.

adjust run level

In the file "/etc/inittab" we have the entry "id:3:initdefault:" which says that 3 is the default run level. The entry "x:5:respawn:/usr/sbin/kdm" starts kdm in run level 5. "respawn" means that kdm is called again after a crash.


To modify the message the console shows if it appears as "Ubuntu feisty (development branch) \n \l" you can modify the text inside the file "/etc/issue".