[ previous ] [ Contents ] [ 1 ] [ 2 ] [ 3 ] [ 4 ] [ 5 ] [ 6 ] [ 7 ] [ 8 ] [ 9 ] [ 10 ] [ 11 ] [ 12 ] [ 13 ] [ 14 ] [ 15 ] [ A ] [ next ]


Debian Reference
Chapter 8 - Debian tips


8.1 Booting the system

See the LDP BootPrompt-HOWTO for detailed information on the boot prompt.


8.1.1 "I forgot the root password!" (1)

It is possible to boot a system and log on to the root account without knowing the root password as long as one has access to the console keyboard. (This assumes there are no password requests from the BIOS or from a boot loader such as lilo that would prevent one from booting the system.)

This is a procedure which requires no external boot disks and no change in BIOS boot settings. Here, "Linux" is the label for booting the Linux kernel in the default Debian install.

At the lilo boot screen, as soon as boot: appears (you must press a shift key at this point on some systems to prevent automatic booting and when lilo uses the framebuffer you have to press TAB to see the options you type), enter:

     boot: Linux init=/bin/sh

This causes the system to boot the kernel and run /bin/sh instead of its standard init. Now you have gained root privileges and a root shell. Since / is currently mounted read-only and many disk partitions have not been mounted yet, you must do the following to have a reasonably functioning system.

     init-2.03# mount -n -o remount,rw /
     init-2.03# mount -avt nonfs,noproc,nosmbfs
     init-2.03# cd /etc
     init-2.03# vi passwd
     init-2.03# vi shadow

(If the second data field in /etc/passwd is "x" for every username, your system uses shadow passwords, and you must edit /etc/shadow.) To disable the root password, edit the second data field in the password file so that it is empty. Now the system can be rebooted and you can log on as root without a password. When booting into runlevel 1, Debian (at least after Potato) requires a password, which some older distributions did not.

It is a good idea to have a minimal editor in /bin/ in case /usr/ is not accessible (see Rescue editors, Section 11.2).

Also consider installing the sash package. When the system becomes unbootable, execute:

     boot: Linux init=/bin/sash

sash serves as an interactive substitute for sh even when /bin/sh is unusable. It's statically linked, and includes many standard utilities as built-ins (type "help" at the prompt for a reference list).


8.1.2 "I forgot the root password!" (2)

Boot from any emergency boot/root disk set. If /dev/hda3 is the original root partition, the following will let one edit the password file just as easily as the above.

     # mkdir fixit
     # mount /dev/hda3 fixit
     # cd fixit/etc
     # vi shadow
     # vi passwd

The advantage of this approach over the previous method is one does not need to know the lilo password (if any). But to use it one must be able to access the BIOS setup to allow the system to boot from floppy disk or CD, if that is not already set.


8.1.3 Cannot boot the system

No problem, even if you didn't bother to make a boot disk during install. If lilo is broken, grab the boot disk from the Debian installation set and boot your system from it. At the boot prompt, assuming the root partition of your Linux installation is on /dev/hda12 and you want runlevel 3, enter:

     boot: rescue root=/dev/hda12 3

Then you are booted into an almost fully functional system using the kernel on the floppy. (There may be minor glitches due to lack of kernel features or modules.)

See also Install a package into an unbootable system, Section 6.3.6 if you have a broken system.

If you need a custom boot floppy, follow readme.txt on the rescue disk.


8.1.4 "Let me disable X on boot!"

Chasing unstable/sid is fun, but buggy xdm, gdm, kdm, or wdm started during the boot process can bite you bad.

First get the root shell by entering the following at the boot prompt:

     boot: Linux vga=normal s

Here, Linux is the label for the kernel image you are booting; "vga=normal" will make sure lilo runs in normal VGA screen, and "s" (or "S") is the parameter passed to init to invoke single-user mode. Enter the root password at the prompt.

There are few ways to disable all the X starting daemons:

Here, number in rc2.d must correspond to the runlevel specified in the /etc/inittab. Also ?dm means that you need to run the command multiple times by substituting it with all of the xdm, gdm, kdm, and wdm.

Only the first one in the list is "the one true way" in Debian. The last one is easy but only works on Debian and requires you to set the display manager again later using dpkg-reconfigure. Others are generic methods to disable daemons.

You can still start X with the startx command from any console shell.


8.1.5 Other boot tricks with the boot prompt

The system can be booted into a particular runlevel and configuration using the lilo boot prompt. Details are given in the BootPrompt-HOWTO (LDP).

If you want to boot the system into runlevel 4, use the following input at the lilo boot prompt.

     boot: Linux 4

If you want to boot the system into normally functioning single-user mode and you know the root password, one of the following examples at the lilo boot prompt will work.

     boot: Linux S
     boot: Linux 1
     boot: Linux -s

If you want to boot the system with less memory than system actually has (say 48MB for a system with 64MB), use this input at the lilo boot prompt:

     boot: Linux mem=48M

Make sure not to specify more than the actual memory size here, otherwise the kernel will crash. If one has more than 64MB of memory, e.g. 128MB, unless one executes mem=128M at the boot prompt or includes a similar append line in /etc/lilo.conf, old kernels and/or a motherboard with an old BIOS will not use memory beyond 64MB.


8.1.6 Setting GRUB boot parameters

GRUB is a new boot manager from the GNU Hurd project and is much more flexible than Lilo but has slightly different handling of boot parameters.

     grub> find /vmlinuz
     grub> root (hd0,0)
     grub> kernel /vmlinuz root=/dev/hda1
     grub> initrd /initrd
     grub> boot

Here, you must be aware of the Hurd device names:

     the Hurd/GRUB       Linux               MS-DOS/Windows
      (fd0)               /dev/fd0            A:
      (hd0,0)             /dev/hda1           C: (usually)
      (hd0,3)             /dev/hda4           F: (usually)
      (hd1,3)             /dev/hdb4           ?

See /usr/share/doc/grub/README.Debian.gz and /usr/share/doc/grub-doc/html/ for details.


8.2 Recording activities


8.2.1 Recording shell activities

System administration involves much more elaborate tasks in a Unix environment than in an ordinary personal computer environment. Make sure to know the most basic means of configuration in case you need to recover from system trouble. X11-based GUI configuration tools look nice and convenient but are often unsuitable in these emergency situations.

In this context, recording shell activities is a good practice, especially as root.

Emacs: Use M-x shell to start recording into a buffer, and use C-x C-w to write the buffer to a file.

Shell: Use the screen command with "^A H" as described in Console switching with screen, Section 8.6.28; or use the script command.

     $ script
     Script started, file is typescript
      ... do whatever ...
      Ctrl-D
     $ col -bx <typescript >savefile
     $ vi savefile

The following can be used instead of script:

     $ bash -i 2>&1 | tee typescript

8.2.2 Recording X activities

If you need to record the graphic image of an X application, including an xterm display, use gimp (GUI). It can capture each window or the whole screen. Alternatives are xwd (xbase-clients), import (imagemagick), and scrot (scrot).


8.3 Copy and archive a whole subdirectory

These copy and archive commands provide basics for the backup of the system and the data. An example of simple backup script is provided as backup in the example scripts.


8.3.1 Basic commands for copying a whole subdirectory

If you need to rearrange file structure, move content including file links by:

     Standard method:
     # cp -a /source/directory /dest/directory # requires GNU cp
     # (cd /source/directory && tar cf - . ) | \
             (cd /dest/directory && tar xvfp - )
     If a hard link is involved, a pedantic method is needed:
     # cd /path/to/old/directory
     # find . -depth -print0 | afio -p -xv -0a /mount/point/of/new/directory
     If remote:
     # (cd /source/directory && tar cf - . ) | \
             ssh user@host.dom (cd /dest/directory && tar xvfp - )
     If there are no linked files:
     # scp -pr user1@host1.dom:/source/directory \
               user2@host2.dom:/dest/directory

The following comparative information on copying a whole subdirectory was presented by Manoj Srivastava srivasta@debian.org to debian-user@lists.debian.org.


8.3.2 cp

Traditionally, cp was not really a candidate for this task since it did not dereference symbolic links, or preserve hard links either. Another thing to consider was sparse files (files with holes).

GNU cp has overcome these limitations; however, on a non-GNU system, cp could still have problems. Also, you can't generate small, portable archives using cp.

     % cp -a . newdir

8.3.3 tar

Tar overcame some of the problems that cp had with symbolic links. However, although cpio handles special files, traditional tar doesn't.

tar's way of handling multiple hard links to a file places only one copy of the link on the tape, but the name attached to that copy is the only one you can use to retrieve the file; cpio's way puts one copy for every link, but you can retrieve it using any of the names.

The tar command changed its option for .bz2 files between Potato and Woody, so use --bzip2 in scripts instead of its short form -I (Potato) or -j (Woody).


8.3.4 pax

The new, POSIX (IEEE Std 1003.2-1992, pages 380–388 (section 4.48) and pages 936–940 (section E.4.48)), all-singing, all-dancing, Portable Archive Interchange utility. pax will read, write, and list the members of an archive file, and will copy directory hierarchies. pax operation is independent of the specific archive format, and supports a wide variety of different archive formats.

pax implementations are still new and wet behind the ears.

     # apt-get install pax
     $ pax -rw -p e . newdir
      or
     $ find . -depth  | pax -rw -p e  newdir

8.3.5 cpio

cpio copies files into or out of a cpio or tar archive. The archive can be another file on the disk, a magnetic tape, or a pipe.

     $ find . -depth -print0 | cpio --null --sparse -pvd new-dir

8.3.6 afio

afio is a better way of dealing with cpio-format archives. It is generally faster than cpio, provides more diverse magnetic tape options and deals somewhat gracefully with input data corruption. It supports multivolume archives during interactive operation. afio can make compressed archives that are much safer than compressed tar or cpio archives. afio is best used as an "archive engine" in a backup script.

     $ find . -depth -print0 | afio -px -0a new-dir

All my backups onto tape use afio.


8.4 Differential backup and data synchronization

Differential backup and data synchronization can be implemented with several methods:

Combination of one of these with the archiving method described in Copy and archive a whole subdirectory, Section 8.3 and the automated regular job described in Schedule activity (cron, at), Section 8.6.27 will make a nice backup system.

I will explain three easy-to-use utilities.


8.4.1 Differential backup with rdiff

rdiff-backup offers nice and simple backup with differential history for any types of files, including symlinks. To back up most of ~/ to /mnt/backup:

     $ rdiff-backup --include ~/tmp/keep --exclude ~/tmp  ~/ /mnt/backup

To restore three-day-old data from this archive to ~/old:

     $ rdiff-backup -r 3D /mnt/backup ~/old

See rdiff-backup(1).


8.4.2 Daily backup with pdumpfs

pdumpfs is a simple daily backup system similar to Plan9's dumpfs which preserves every daily snapshot. You can access the past snapshots at any time for retrieving a certain day's file. Let's backup your home directory with pdumpfs and cron!

pdumpfs constructs the snapshot YYYY/MM/DD in the destination directory. All source files are copied to the snapshot directory when pdumpfs is run for the first time. On and after the second time, pdumpfs copies only updated or newly created files and stores unchanged files as hard links to the files of the previous day's snapshot in order to save disk space.

     $ pdumpfs src-dir dest-dir [dest-basename]

See pdumpfs(8).


8.4.3 Regular differential backup with RCS

Changetrack will record changes to the text-based configuration files in RCS archives regularly. See changetrack(1).

     # apt-get install changetrack
     # vi changetrack.conf

8.5 System freeze recovery


8.5.1 Kill a process

Run top to see what process is acting funny. Press `P' to sort by CPU usage, `M' to sort by memory, and `k' to kill a process. Alternatively, BSD-style ps aux | less or System-V-style ps -efH | less may be used. The System-V-style syntax displays parent process IDs (PPID) which can be used for killing zombie (defunct) children.

Use kill to kill (or send a signal to) a process by process ID, killall to do the same by process command name. Frequently used signals:

      1: HUP,  restart daemon
     15: TERM, normal kill
      9: KILL, kill hard

8.5.2 Alt-SysRq

Insurance against system malfunction is provided by the kernel compile option "Magic SysRq key". Pressing Alt-SysRq on an i386, followed by one of the keys r 0 k e i s u b, does the magic.

Un`r'aw restores the keyboard after things like X crashes. Changing the console loglevel to `0' reduces error messages. sa`k' (system attention key) kills all processes on the current virtual console. t`e'rminate kills all processes on the current terminal except init. k`i'll kills all processes except init.

`S'ync, `u'mount, and re`b'oot are for getting out of really bad situations.

Debian default installation kernels are not compiled with this option at the time this document is written. Recompile the kernel to activate this function. Detailed information is in /usr/share/doc/kernel-doc-version/Documentation/sysrq.txt.gz or /usr/src/kernel-version/Documentation/sysrq.txt.gz.


8.6 Nifty little commands to remember


8.6.1 Pager

less is the default pager (file content browser). Hit `h' for help. It can do much more than more. less can be supercharged by executing eval $(lesspipe) or eval $(lessfile) in the shell startup script. See more in /usr/share/doc/lessf/LESSOPEN. The -R option allows raw character output and enables ANSI color escape sequences. See less(1).

w3m may be a useful alternative pager for some code systems (EUC).


8.6.2 Free memory

free and top give good information on memory resources. Do not worry about the size of "used" in the "Mem:" line, but read the one under it (38792 in the example below).

     $ free -k # for 256MB machine
                  total       used       free     shared    buffers cached
     Mem:        257136     230456      26680      45736     116136 75528
     -/+ buffers/cache:      38792     218344
     Swap:       264996          0     264996

The exact amount of physical memory can be confirmed by grep '^Memory' /var/log/dmesg, which in this case gives "Memory: 256984k/262144k available (1652k kernel code, 412k reserved, 2944k data, 152k init)".

     Total         = 262144k = 256M (1k=1024, 1M=1024k)
     Free to dmesg = 256984k = Total - kernel - reserved - data - init
     Free to shell = 257136k = Total - kernel - reserved - data

About 5MB is not usable by the system because the kernel uses it.


8.6.3 Set time (BIOS)

     # date MMDDhhmmCCYY
     # hwclock --utc --systohc
     # hwclock --show

This will set system and hardware time to MM/DD hh:mm, CCYY. Times are displayed in local time but hardware time uses UTC.

If the hardware (BIOS) time is set to GMT, change the setting to UTC=yes in the /etc/default/rcS.


8.6.4 Set time (NTP)

Reference: Managing Accurate Date and Time HOWTO.


8.6.4.1 Set time with permanent Internet connection

Set system clock to the correct time automatically via a remote server:

     # ntpdate server

This is good to have in /etc/cron.daily/ if your system has a permanent Internet connection.


8.6.4.2 Set time with sporadic Internet connection

Use the chrony package.


8.6.5 How to control console features such as the screensaver

For disabling the screensaver, use following commands.

In the Linux console:

     # setterm -powersave off

Start the kon2 (kanji) console with:

     # kon -SaveTime 0

While running X:

     # xset s off
      or
     # xset -dpms
      or
     # xscreensaver-command -prefs

Read the corresponding manpages for controlling other console features. See also stty(1) for changing and printing terminal line settings.


8.6.6 Search administrative database

Glibc offers getent(1) for searching entries from administrative databases, i.e., passwd, group, hosts, services, protocols, or networks.

     getent database [key ...]

8.6.7 Disable sound (beep)

One can always unplug the PC speaker. ;-) For the Bash shell:

     echo "set bell-style none">> ~/.inputrc

8.6.8 Error messages on the console screen

In order to quiet on-screen error messages, the first place to check is /etc/init.d/klogd. Set KLOGD="-c 3" in this script and run /etc/init.d/klogd restart. An alternative method is to run dmesg -n3.

Here error levels mean:

If one particular useless error message bothers you a lot, consider making a trivial kernel patch like shutup-abit-bp6 (available in the examples subdirectory).

Another place to look may be /etc/syslog.conf; check to see whether any messages are logged to a console device.


8.6.9 Set console to the correct type

Console screens in Unix-like systems are usually accessed using (n)curses library routines. These give the user a terminal-independent method of updating character screens with reasonable optimization. See ncurses(3X) and terminfo(5).

On a Debian system, there are quite a lot of predefined entries:

     $ toe | less                  # all entries
     $ toe /etc/terminfo/ | less   # user reconfigurable entries

Export your selection as environment variable TERM.

If the terminfo entry for xterm doesn't work with a non-Debian xterm, change your terminal type from "xterm" to one of the feature-limited versions such as "xterm-r6" when you log in to a Debian system remotely. See /usr/share/doc/libncurses5/FAQ for more. "dumb" is the lowest common denominator for terminfo.


8.6.10 Get the console back to a sane state

When the screen goes berserk after cat some-binary-file (you may not be able to see the command echoed as you type):

     $ reset

8.6.11 Convert a text file from DOS to Unix style

Convert a DOS text file (end-of-line = ^M^J) to a Unix text file (end-of-line = ^J).

     # apt-get install sysutils
     $ dos2unix dosfile

8.6.12 Convert a text file with recode

Following will convert text files between DOS, Mac, and Unix line ending styles:

     $ recode /cl../cr <dos.txt >mac.txt
     $ recode /cr.. <mac.txt >unix.txt
     $ recode ../cl <unix.txt >dos.txt

Free recode converts files between various character sets and surfaces with:

     $ recode charset1/surface1..charset2/surface2 \
       <input.txt >output.txt

Common character sets used are (see also Introduction to locales, Section 9.7.3) [37] :

Common surfaces used are [38] :

For more, see pertinent description in the info recode.

There are also more specialized conversion tools:


8.6.13 Regular-expression substitution

Replace all instances of FROM_REGEX with TO_TEXT in all of the files FILES ...:

     $ perl -i -p -e 's/FROM_REGEX/TO_TEXT/g;' FILES ...

-i is for "in-place editing", -p is for "implicit loop over FILES ...". If the substitution is complex, you can make recovery from errors easier by using the parameter -i.bak instead of -i; this will keep each original file, adding .bak as a file extension.


8.6.14 Edit a file in place using a script

The following script will remove lines 5–10 and lines 16–20 in place.

     #!/bin/bash
     ed $1 <<EOF
     16,20d
     5,10d
     w
     q
     EOF

Here, ed commands are the same as vi command-mode commands. Editing from the back of file makes it easy for scripting.


8.6.15 Extract differences and merge updates for source files

Following one of these procedures will extract differences between two source files and create unified diff files file.patch0 or file.patch1 depending on the file location:

     $ diff -u file.old file.new > file.patch0
     $ diff -u old/file new/file > file.patch1

The diff file (alternatively called patch file) is used to send a program update. The receiving party will apply this update to another file by:

     $ patch -p0 file < file.patch0
     $ patch -p1 file < file.patch1

If you have three versions of source code, you can merge them more effectively using diff3:

     $ diff3 -m file.mine file.old file.yours > file

8.6.16 Convert a large file into small files

     $ split -b 650m file   # split file into 650MB chunks
     $ cat x* >largefile    # merge files into 1 large file

8.6.17 Extract data from text file table

Let's consider a text file called DPL in which all previous Debian project leader's names and their initiation days are listed in a space-separated format.

     Ian     Murdock   August  1993
     Bruce   Perens    April   1996
     Ian     Jackson   January 1998
     Wichert Akkerman  January 1999
     Ben     Collins   April   2001
     Bdale   Garbee    April   2002
     Martin  Michlmayr March   2003

Awk is frequently used to extract data from these types of files.

     $ awk '{ print $3 }' <DPL                   # month started
     August
     April
     January
     January
     April
     April
     March
     $ awk '($1=="Ian") { print }' <DPL          # DPL called Ian
     Ian     Murdock   August  1993
     Ian     Jackson   January 1998
     $ awk '($2=="Perens") { print $3,$4 }' <DPL # When Perens started
     April 1996

Shells such as Bash can be also used to parse this kind of file:

     $ while read first last month year; do 
         echo $month
       done <DPL
     ... same output as the first Awk example

Here, read built-in command uses the characters in $IFS (internal field separators) to split lines into words.

If you change IFS to ":", you can parse /etc/passwd with shell nicely:

     $ oldIFS="$IFS"   # save old value
     $ IFS=":"
     $ while read user password uid gid rest_of_line; do
         if [ "$user" = "osamu" ]; then 
           echo "$user's ID is $uid"
         fi
       done < /etc/passwd
     osamu's ID is 1001
     $ IFS="$oldIFS"   # restore old value

(If Awk is used to do the equivalent, use FS=":" to set the field separator.)

IFS is also used by the shell to split results of parameter expansion, command substitution, and arithmetic expansion. These do not occur within double or single quoted words. The default value of IFS is <space>, <tab>, and <newline> combined.

Be careful about using this shell IFS tricks. Strange things may happen, when shell interprets some parts of the script as its input.

     $ IFS=":,"                        # use ":" and "," as IFS
     $ echo IFS=$IFS,   IFS="$IFS"     # echo is a Bash built-in
     IFS=  , IFS=:,
     $ date -R                         # just a command output
     Sat, 23 Aug 2003 08:30:15 +0200
     $ echo $(date -R)                 # sub shell --> input to main shell
     Sat  23 Aug 2003 08 30 36 +0200
     $ unset IFS                       # reset IFS to the default
     $ echo $(date -R)
     Sat, 23 Aug 2003 08:30:50 +0200

8.6.18 Script snippets for piping commands

The following scripts will do nice things as a part of a pipe.

     find /usr | egrep -v "/usr/var|/usr/tmp|/usr/local"
                          # find all files in /usr excluding some files
     xargs -n 1 command   # run command for all items from stdin
     xargs -n 1 echo |    # split white-space-separated items into lines
     xargs echo      |    # merge all lines into a line
     grep -e pattern|     # extract lines containing pattern
     cut -d: -f3 -|
             # extract third field separated by : (passwd file etc.)
     awk '{ print $3 }' | # extract third field separated by whitespaces
     awk -F'\t' '{ print $3 }' |
             # extract third field separated by tab
     col -bx |            # remove backspace and expand tabs to spaces
     expand -|            # expand tabs
     sort -u|             # sort and remove duplicates
     
     tr '\n' ' '|         # concatenate lines into one line
     tr '\r' ''|          # remove CR
     tr 'A-Z' 'a-z'|      # convert uppercase to lowercase
     sed 's/^/# /'|       # make each line a comment
     sed 's/\.ext//g'|    # remove .ext
     sed  -n -e 2p|       # print the second line 
     head -n 2 -|         # print the first 2 lines
     tail -n 2 -|         # print the last 2 lines

8.6.19 Script snippets for looping over each file

The following ways of looping over each file matching *.ext ensures proper handling of funny file names such as ones with spaces and performs equivalent process:


8.6.20 Perl short script madness

Although any Awk scripts can be automatically rewritten in Perl using a2p(1), one-liner Awk scripts are best converted to one-liner perl scripts manually. For example

     awk '($2=="1957") { print $3 }' |

is equivalent to any one of the following lines:

     perl -ne '@f=split; if ($f[1] eq "1957") { print "$f[2]\n"}' |
     perl -ne 'if ((@f=split)[1] eq "1957") { print "$f[2]\n"}' |
     perl -ne '@f=split; print $f[2] if ( $f[1]==1957 )' |
     perl -lane 'print $F[2] if $F[1] eq "1957"' |

Since all the whitespace in the arguments to perl in the line above can be removed, and taking advantage of the automatic conversions between numbers and strings in Perl:

     perl -lane 'print$F[2]if$F[1]eq+1957' |

See perlrun(1) for the command-line options. For more crazy Perl scripts, http://perlgolf.sourceforge.net may be interesting.


8.6.21 Get text or a mailing list archive from a web page

The following will read a web page into a text file. Very useful when copying configurations off the Web.

     $ lynx -dump http://www.remote-site.com/help-info.html >textfile

links and w3m can be used here, too, with slight differences in rendering.

If this is a mailing list archive, use munpack to obtain mime contents from text.


8.6.22 Pretty print a web page

The following will print a web page into a PostScript file/printer.

     $ apt-get install html2ps
     $ html2ps URL | lpr

See lpr/lpd, Section 3.6.1. Also check a2ps and mpage packages for creating PostScript files.


8.6.23 Pretty print a manual page

The following will print a manual page into a PostScript file/printer.

     $ man -Tps some-manpage | lpr
     $ man -Tps some-manpage | mpage -2 | lpr

8.6.24 Merge two PostScript or PDF files

You can merge two PostScript or PDF files.

     $ gs -q -dNOPAUSE -dBATCH -sDEVICE=pswrite \
       -sOutputFile=bla.ps -f foo1.ps foo2.ps
     $ gs -q -dNOPAUSE -dBATCH -sDEVICE=pdfwrite \
       -sOutputFile=bla.pdf -f foo1.pdf foo2.pdf

8.6.25 Time a command

Display time used by a process.

     # time some-command >/dev/null
     real    0m0.035s       # time on wall clock (elapsed real time)
     user    0m0.000s       # time in user mode
     sys     0m0.020s       # time in kernel mode

8.6.26 nice command

Use nice (from the GNU shellutils package) to set a command's nice value when starting. renice (bsdutils) and top can renice a process. A nice value of 19 represents the slowest (lowest priority) process; negative values are "not-nice", with -20 being a very fast (high priority) process. Only the superuser can set negative nice values.

     # nice  -19 top                                         # very nice
     # nice --20 cdrecord -v -eject speed=2 dev=0,0 disk.img # very fast

Sometimes an extreme nice value does more harm than good to the system. Use this command carefully.


8.6.27 Schedule activity (cron, at)

Use cron and at to schedule tasks under Linux. See at(1), crontab(5), crontab(8).

Run the command crontab -e to create or edit a crontab file to set up regularly scheduled events. Example of a crontab file:

     # use /bin/sh to run commands, no matter what /etc/passwd says
     SHELL=/bin/sh
     # mail any output to `paul', no matter whose crontab this is
     MAILTO=paul
     # Min Hour DayOfMonth Month DayOfWeek command (Day... are OR'ed)
     # run at 00:05, every day
     5  0  *  * *   $HOME/bin/daily.job >> $HOME/tmp/out 2>&1
     # run at 14:15 on the first of every month -- output mailed to paul
     15 14 1  * *   $HOME/bin/monthly
     # run at 22:00 on weekdays(1-5), annoy Joe. % for newline, last % for cc:
     0 22 *   * 1-5 mail -s "It's 10pm" joe%Joe,%%Where are your kids?%.%%
     23 */2 1 2 *   echo "run 23 minutes after 0am, 2am, 4am ..., on Feb 1"
     5  4 *   * sun echo "run at 04:05 every sunday"
     # run at 03:40 on the first Monday of each month
     40 3 1-7 * *   [ "$(date +%a)" == "Mon" ] && command -args

Run the at command to schedule a one-time job:

     $ echo 'command -args'| at 3:40 monday

8.6.28 Console switching with screen

The screen program allows you to run multiple virtual terminals, each with its own interactive shell, on a single physical terminal or terminal emulation window. Even if you use Linux virtual consoles or multiple xterm windows, it is worth exploring screen for its rich feature set, which includes


8.6.28.1 Remote access scenario

If you frequently log on to a Linux machine from a remote terminal or using a VT100 terminal program, screen will make your life much easier with the detach feature.

  • You are logged in via a dialup connection, and are running a complex screen session with editors and other programs open in several windows.

  • Suddenly you need to leave your terminal, but you don't want to lose your work by hanging up.

  • Simply type ^A d to detach the session, then log out. (Or, even quicker, type ^A DD to have screen detach and log you out itself.)

  • When you log on again later, enter the command screen -r, and screen will magically reattach all the windows you had open.


  • 8.6.28.2 Typical screen commands

    Once you start screen, all keyboard input is sent to your current window except for the command keystroke, by default ^A. All screen commands are entered by typing ^A plus a single key [plus any parameters]. Useful commands:

         ^A ?     show a help screen (display key bindings)
         ^A c     create a new window and switch to it
         ^A n     go to next window
         ^A p     go to previous window
         ^A 0     go to window number 0
         ^A w     show a list of windows
         ^A a     send a Ctrl-A to current window as keyboard input
         ^A h     write a hardcopy of current window to file 
         ^A H     begin/end logging current window to file
         ^A ^X    lock the terminal (password protected)
         ^A d     detach screen session from the terminal
         ^A DD    detach screen session and log out
    

    This is only a small subset of screen's commands and features. If there's something you want screen to be able to do, chances are it can! See screen(1) for details.


    8.6.28.3 Backspace and/or Ctrl-H in screen session

    If you find that backspace and/or Ctrl-H do not work properly when you are running screen, edit /etc/screenrc, find the line reading

         bindkey -k kb stuff "\177"
    

    and comment it out (i.e., add "#" as the first character).


    8.6.28.4 Equivalent program to screen for X

    Check out xmove. See xmove(1).


    8.6.29 Network testing basics

    Install netkit-ping, traceroute, dnsutils, ipchains (for 2.2 kernel), iptables (for 2.4 kernel), and net-tools packages and:

         $ ping yahoo.com            # check Internet connection
         $ traceroute yahoo.com      # trace IP packets
         $ ifconfig                  # check host config
         $ route -n                  # check routing config
         $ dig [@dns-server.com] host.dom [{a|mx|any}] |less
               # check host.dom DNS records by dns-server.com 
               # for a {a|mx|any} record
         $ ipchains -L -n |less      # check packet filter (2.2 kernel)
         $ iptables -L -n |less      # check packet filter (2.4 kernel)
         $ netstat -a                # find all open ports
         $ netstat -l --inet         # find listening ports
         $ netstat -ln --tcp         # find listening TCP ports (numeric)
    

    8.6.30 Flush mail from local spool

    To flush mail from the local spool:

         # exim4 -q    # flush waiting mail
         # exim4 -qf   # flush all mail
         # exim4 -qff  # flush even frozen mail
    

    -qff may be better as an option in the /etc/ppp/ip-up.d/exim script. For Woody and older distributions, replace exim4 with exim.


    8.6.31 Remove frozen mail from local spool

    To remove frozen mail from the local spool with a delivery error message:

         # exim4 -Mg `mailq | grep frozen | awk '{ print $3 }'`
    

    For Woody and older distributions, replace exim4 with exim.


    8.6.32 Redeliver mbox contents

    You need to manually deliver mails to the sorted mailboxes in your home directory from /var/mail/username if your home directory became full and procmail failed. After making disk space in the home directory, run:

         # /etc/init.d/exim4 stop
         # formail -s procmail </var/mail/username
         # /etc/init.d/exim4 start
    

    For Woody and older distributions, replace exim4 with exim.


    8.6.33 Clear file contents

    In order to clear the contents of a file such as a logfile, do not use rm to delete the file and then create a new empty file, because the file may still be accessed in the interval between commands. The following is the safe way to clear the contents of the file.

         $ :>file-to-be-cleared
    

    8.6.34 Dummy files

    The following commands will create dummy or empty files:

         $ dd if=/dev/zero    of=filename bs=1k count=5 # 5KB of zero content
         $ dd if=/dev/urandom of=filename bs=1M count=7 # 7MB of random content
         $ touch filename #  create 0B file (if file exists, updates mtime)
    

    For example, the following commands executed from the shell of the Debian boot floppy will erase all the content of the hard disk /dev/hda completely for most practical uses.

         # dd if=/dev/urandom of=/dev/hda; dd if=/dev/zero of=/dev/hda
    

    8.6.35 chroot

    The chroot program, chroot(8), enables us to run different instances of the GNU/Linux environment on a single system simultaneously without rebooting.

    One may also run a resource hungry program such as apt-get or dselect under the chroot of a fast host machine while NFS-mounting a slow satellite machine to the host as r/w and the chroot point being the mount point of the satellite machine.


    8.6.35.1 Run a different Debian distribution with chroot

    A chroot Debian environment can easily be created by the debootstrap command in Sarge. For post-Sarge distributions, you may use cdebootstrap command instead with appropriate option. For example, to create a Sid chroot on /sid-root while having fast Internet access:

         main # cd /; mkdir /sid-root
         main # debootstrap sid /sid-root http://ftp.debian.org/debian/
         ... watch it download the whole system
         main # echo "proc /sid-root/proc proc none 0 0" >> /etc/fstab
         main # mount /sid-root/proc
         main # mount /dev/ /sid-root/dev -o bind
         main # cp /etc/hosts /sid-root/etc/hosts
         main # chroot /sid-root /bin/bash
         chroot # cd /dev; /sbin/MAKEDEV generic; cd -
         chroot # apt-setup # set-up /etc/apt/sources.list
         chroot # vi /etc/apt/sources.list # point the source to unstable
         chroot # dselect  # you may use aptitude, install mc and vim :-)
    

    At this point you should have a fully working Debian system, where you can play around without fear of affecting your main Debian installation.

    This debootstrap trick can also be used to install Debian to a system without using a Debian install disk, but instead one for another GNU/Linux distribution. See http://www.debian.org/releases/stable/i386/apcs04.


    8.6.35.2 Setting up login for chroot

    Typing chroot /sid-root /bin/bash is easy, but it retains all sorts of environment variables that you may not want, and has other issues. A much better approach is to run another login process on a separate virtual terminal where you can log in to the chroot directly.

    Since on default Debian systems tty1 to tty6 run Linux consoles and tty7 runs the X Window System, let's set up tty8 for a chrooted console as an example. After creating a chroot system as described in Run a different Debian distribution with chroot, Section 8.6.35.1, type from the root shell of the main system:

         main # echo "8:23:respawn:/usr/sbin/chroot /sid-root "\
                "/sbin/getty 38400 tty8" >> /etc/inittab
         main # init q    # reload init
    

    8.6.35.3 Setting up X for chroot

    You want to run the latest X and GNOME safely in your chroot? That's entirely possible! The following example will make GDM run on virtual terminal vt9.

    First install a chroot system using the method described in Run a different Debian distribution with chroot, Section 8.6.35.1. From the root of the main system, copy key configuration files to the chroot system.

         main # cp /etc/X11/XF86Config-4 /sid-root/etc/X11/XF86Config-4
         main # chroot /sid-root # or use chroot console
         chroot # cd /dev; /sbin/MAKEDEV generic; cd -
         chroot # apt-get install gdm gnome x-window-system
         chroot # vi /etc/gdm/gdm.conf # do s/vt7/vt9/ in [servers] section
         chroot # /etc/init.d/gdm start
    

    Here, /etc/gdm/gdm.conf was edited to change the first virtual console from vt7 to vt9.

    Now you can easily switch back and forth between full X environments in your chroot and your main system just by switching between Linux virtual terminals; e.g. by using Ctrl-Alt-F7 and Ctrl-Alt-F9. Have fun!

    [FIXME] Add a comment and link to the init script of the chrooted gdm.


    8.6.35.4 Run other distributions with chroot

    A chroot environment for another Linux distribution can easily be created. You install a system into separate partitions using the installer of the other distribution. If its root partition is in /dev/hda9:

         main # cd /; mkdir /other-dist
         main # mount -t ext3 /dev/hda9 /other-dist
         main # chroot /other-dist /bin/bash
    

    Then proceed as in Run a different Debian distribution with chroot, Section 8.6.35.1, Setting up login for chroot, Section 8.6.35.2, and Setting up X for chroot, Section 8.6.35.3.


    8.6.35.5 Build a package with chroot

    There is a more specialized chroot package, pbuilder, which constructs a chroot system and builds a package inside the chroot. It is an ideal system to use to check that a package's build-dependencies are correct, and to be sure that unnecessary and wrong build dependencies will not exist in the resulting package.


    8.6.36 How to check hard links

    You can check whether two files are the same file with two hard links by:

          
         $ ls -li file1 file2
    

    8.6.37 mount hard disk image file

    If file.img contains an image of hard disk contents and the original hard disk had a disk configuration which gives xxxx = (bytes/sector) * (sectors/cylinder), then the following will mount it to /mnt:

         # mount -o loop,offset=xxxx file.img /mnt
    

    Note that most hard disks have 512 bytes/sector.


    8.6.38 Samba

    Basics of getting files from Windows:

         # mount -t smbfs -o username=myname,uid=my_uid,gid=my_gid \
                 //server/share /mnt/smb  # mount Windows files to Linux
         # smbmount //server/share /mnt/smb \
                 -o "username=myname,uid=my_uid,gid=my_gid"
         # smbclient -L 192.168.1.2 # list the shares on a computer
    

    Samba neighbors can be checked from Linux using:

         # smbclient -N -L ip_address_of_your_PC | less
         # nmblookup -T "*"
    

    8.6.39 Utilities for foreign filesystems

    Many foreign filesystems have Linux kernel support, and can thus be accessed simply by mounting the devices containing the filesystems. For certain filesystems, there are also a few specialized tools to access the filesystems without mounting the devices. This is accomplished with user-space programs so that kernel filesystem support is not needed.

    In order to create and check an MS-DOS FAT filesystem, dosfstools is useful.


    8.7 Typical mistakes to be noted

    Here are few examples of dangerous actions. The negative impacts will be enhanced if you are using privileged account: root.


    8.7.1 rm -rf .*

    The use of wild card file name in command line arguments such as "rm -rf .*" may cause dangerous result, since ".*" expands to include "." and "..". Fortunately for the current verion of "rm" command in the Debian distribution, it checks sanity of the argument file names and refuses to remove "." and "..". But this is not always the case. Try following to see how the wild card file names work.


    8.7.2 rm /etc/passwd

    Loss of some important files such as /etc/passwd through your stupidity is tough. The Debian system makes regular backups of them in /var/backups/. When you restore these files, you may manually have to set the proper permissions.

         # cp /var/backups/passwd /etc/passwd
         # chmod 644 /etc/passwd
    

    See also Recover package selection data, Section 6.3.4.


    [ previous ] [ Contents ] [ 1 ] [ 2 ] [ 3 ] [ 4 ] [ 5 ] [ 6 ] [ 7 ] [ 8 ] [ 9 ] [ 10 ] [ 11 ] [ 12 ] [ 13 ] [ 14 ] [ 15 ] [ A ] [ next ]


    Debian Reference

    CVS, Thu Jan 18 11:52:15 UTC 2007

    Osamu Aoki osamu#at#debian.org
    Authors, Section A.1