Inizio  >  Docs  >  Linux  >  How to  >  backup  
backupbackup

Index

DD Command

 

 

 

 

 

duplicazione a caldo

http://nelregnodilinux.blogspot.com/2010/08/duplicazione-caldo-di-un-sistema-linux.html

 

 

http://www.linuxquestions.org/questions/linux-newbie-8/learn-the-dd-command-362506/

 

 

The basic command is structured as follows:

 

dd if=<source> of=<target> bs=<byte size>("USUALLY" some power of 2, not less than 512 bytes(ie, 512, 1024, 2048, 4096, 8192, 16384, but can be ANY reasonable number.) skip= seek= conv=<conversion>.[/b]

Source is the data being read. Target is where the data gets written. Warning!! If you reverse the source and target, you can wipe out a lot of data. This feature has inspired the nickname "dd" Data Destroyer. Warning!! Caution should be observed when using dd to copy encrypted partitions. How to make a swap file, or another swapfile on a running system:

 

dd if=/dev/zero of=/swapspace bs=4k count=250000
mkswap /swapspace
swapon /swapspace

This can solve out of memory issues due to memory leaks on servers that cannot easily be rebooted. New How to pick proper block size:

 

time dd if=/dev/zero bs=1024 count=1000000 of=/home/sam/1Gb.file
time dd if=/dev/zero bs=2048 count=500000 of=/home/sam/1Gb.file
time dd if=/dev/zero bs=4096 count=250000 of=/home/sam/1Gb.file
time dd if=/dev/zero bs=8192 count=125000 of=/home/sam/1Gb.file

This method can also be used as a drive benchmark, to find strengths and weaknesses in hard drives: Read:

 

time dd if=/home/sam/1Gb.file bs=64k | dd of=/dev/null

Write:

 

time dd if=/dev/zero bs=1024 count=1000000 of=/home/sam/1Gb.file

The output looks like this:

 

1000000+0 records in
1000000+0 records out

real 2m17.951s
user 0m0.930s
sys 0m25.160s

Your output may have different figures. The format is what I'm referring to. If you play with 'bs=' and 'count=', always having them multiply out to the same figure, you can calculate bytes/second like this: 1Gb/total seconds = Gb/s. You only use the 'real' timing figure from above. You can get more realistic results using a 3Gb file. How to rejuvenate a hard drive This will sometimes cure input/output errors experienced when using dd. Over time the data on a drive, especially a drive that hasn't been used for a year or two grows into larger magnetic flux points than were originally recorded. This makes it hard for the drive heads to decipher these magnetic flux points. This results in read/write errors. Also, sometimes sector 1 goes bad resulting in a useless drive. All you need to do is:

 

dd if=/dev/sda of=/dev/sda

to rejuvenate the drive. The process rewrites all the data on the drive in nice tight magnetic patterns that can then be read properly. The procedure is perfectly safe, and saves one a lot of money on HDDs. Examples: Copy one hard disk partition to another hard disk:

 

dd if=/dev/sda2 of=/dev/sdb2 bs=4096 conv=notrunc,noerror

sda2 and sdb2 are partitions. You want to copy sda2 to sdb2. If sdb2 doesn't exist, dd will start at the beginning of the disk, and create it. Be careful with order of if and of. You can write a blank disk to a good disk if you get confused. The only difference between a big partition and a small partition, besides size, is the partition table. If you are copying sda to sdb, an entire drive with a single partition, sdb being smaller than sda, then you have to do:

 

dd if=/dev/sda skip=2 of=/dev/sdb seek=2 bs=4k conv=noerror

Skip skips input blocks at the beginning of the media (sda). Seek skips over so many blocks on the output media before writing (sdb). By doing this, you leave the first 4k bytes on each drive untouched. You don't want to tell a drive it is bigger than it really is by writing a partition table from a larger drive to a smaller drive. The first 63 sectors of a drive are empty, except sector 1, the MBR. If you are copying a smaller partition to a larger one, the larger partition will read the correct size with:

 

fdisk -l

, but not with:

 

df -h

This is because fdisk reads the partition table and df reads the format info. If you dd a smaller partition to a larger one the larger one will now be formatted the same as the smaller one and there won't be any space left on the drive. The way around this is to build a partition image file.

 

dd if=smaller_partition of=/home/sam/smaller_partition.img

Mount the image like a drive using:

 

mount -o loop /home/sam/smaller_partition.img /mnt/directory
cd /mnt/directory
cp -r * /mnt/larger_partition_already_partitioned_and_formatted_to_the_size_you_want

Cp has an -r switch for recursive copy. Now, if you are copying sda3 to sda2, this is different. What you want to do is this:

 

dd if=/dev/sda3 of=/dev/sda2 bs=4096 conv=notrunc,noerror

The very last part of the drive is usually zeroes. So, if you have room for the data, and the zeroes get truncated that's ok. Make an iso image of a CD:

 

dd if=/dev/hdc of=/home/sam/mycd.iso bs=2048 conv=notrunc

This copies sector for sector. The result will be a hard disk image file of the CD. You can mount the image with:

 

mkdir /mnt/mycd

, this line in fstab:

 

/home/sam/mycd.iso /mnt/mycd iso9660 rw,user,noauto 0 0

Save fstab,

 

"mount -o loop /mnt/mycd".

Then the file system will be viewable as files and directories in the directory

 

/mnt/mycd

Copy a floppy disk:

 

dd if=/dev/fd0 of=/home/sam/floppy.image bs=2x80x18b conv=notrunc

or

 

dd if=/dev/fd0 of=/home/sam/floppy.image conv=notrunc 

The 18b specifies 18 sectors of 512 bytes, the 2x multiplies the sector size by the number of heads, and the 80x is for the cylinders--a total of 1474560 bytes. This issues a single 1474560-byte read request to /dev/fd0 and a single 1474560 write request to

 

/home/sam/floppy.image

. This makes a hard drive image of the floppy, with bootable info intact. The second example uses default "bs=" of 512, which is the sector size of a floppy.If you're concerned about spies with superconducting quantum-interference detectors you can always add a "for" loop for government level secure disk erasure: copy and paste the following two lines into a text editor.

 

#!/bin/bash 
for n in `seq 7`; do dd if=/dev/urandom of=/dev/sda bs=8b conv=notrunc; done

Now you have a shell script for seven passes of random characters over the whole disk Do:

 

chmod a+x <shellscriptfile> 

to make it executable. To make a bootable USB thumb drive: Download 50 MB Debian based distro here: sourceforge.net/projects/insert/ Plug in the thumb drive to a USB port. Do:

 

dmesg | tail

Look where the new drive is, sdb, or something similar. Do:

 

dd if=/home/sam/insert.iso of=/dev/sdb ibs=4b obs=1b conv=notrunc,noerror 

Now set the BIOS to USB boot, plug in the thumb drive, and boot the machine. Copy just the MBR and boot sector of a floppy to hard drive image:

 

dd if=/dev/fd0 of=/home/sam/MBRboot.image bs=512 count=2 

This copies the first 2 sectors of the floppy. Cloning an entire hard disk:[/color]

 

dd if=/dev/sda of=/dev/sdb conv=notrunc,noerror

In this example, sda is the source. Sdb is the target. Do not reverse the intended source and target. Surprisingly many people do. Notrunc means 'do not truncate the output file'. Noerror means to keep going if there is an error. Normally dd stops at any error. Copy MBR only of a hard drive: [/color]

 

dd if=/dev/sda of=/home/sam/MBR.image bs=446 count=1 

This will copy the first 446 bytes of the hard drive to a file. If you haven't already guessed, reversing the objects of if and of, in the dd command line reverses the direction of the write. Wipe a hard drive of all data (you would want to boot from a cd to do this)[/color] www.efense.com/helix is a good boot cd. The helix boot environment contains the DoD version of dd called dcfldd. It works the same way, but is has a progress bar.

 

dcfldd if=/dev/zero of=/dev/sda conv=notrunc 

This is useful for making the drive almost like new. Most drive have 0x0000ffh written to every sector from the factory. Overwrite all the free space on a partition (deleted files you don't want recovered)[/color]

 

dd if=/dev/urandom > fileconsumingallfreespace

When dd says no room left on device, all the free space has been overwritten with random characters. Then, delete the big file with

 

rm

. To view your virtual memory:

 

dd if=/proc/kcore | hexdump -C | less

use PgUp, PgDn, up arrow, down arrow to navigate in less. Less is my favorite editor. Or I should say it would be my favorite editor if it allowed editing. What filesystems are installed:

 

dd if=/proc/filesystems | hexdump -C | less 

all loaded modules:

 

dd if=/proc/kallsyms | hexdump -C | less 

interrupt table:

 

dd if=/proc/interrupts | hexdump -C | less

How many seconds has the system been up:

 

dd if=/proc/uptime | hexdump -C | less 

partitions and sizes in kb:

 

dd if=/proc/partitions | hexdump -C | less 

Memory stats:

 

dd if=/proc/meminfo | hexdump -C | less 

I put two identical drives in every one of my machines. Before I do anything which might be disasterous, I do:

 

dcfldd if=/dev/sda of=/dev/sdb bs=4096 conv=notrunc,noerror 

and copy my present working sda drive system to the sdb drive. If I wreck the installation on sda, I just boot with the helix cd and:

 

dcfldd if=/dev/sdb of=/dev/sda bs=4096 conv=notrunc,noerror 

and I get everything back exactly the same as before whatever boneheaded thing I was trying to do didn't work. You can really, really learn linux this way, because you absolutely can't wreck what you have an exact copy of. You also might consider making the root partition separate from /home, and make /home big enough to hold the root partition, plus more. Then you can do:

 

dd if=/dev/sda2 (root) of /home/sam/root.img bs=4096 conv=notrunc,noerror

To make a backup of root, and :

 

dd if /home/sam/root.img of=/dev/sda2 (root) bs=4096 conv=notrunc,noerror

To write the image of root back to the root partition if you messed up and can't launch the X server anymore, or edited /etc/fstab and can't figure out what you did wrong. It only takes a few minutes to restore a 15 GB root partition from an image file. To make a file of 100 random bytes:

 

dd if=/dev/urandom of=/home/sam/myrandom bs=100 count=1

/dev/random produces only as many random bits as the entropy pool contains. This yields quality randomness for kryptographic keys. If more random bytes are required, the process stops until the entropy pool is refilled (waggling your mouse helps). /dev/urandom does not have this restriction. If the user demands more bits than currently in the entropy pool, it produces them using a pseudo random number generator. Here, /dev/urandom is the Linux random byte device. myrandom is a file. Write random data over a file before deleting it: first do an

 

ls -l

to find filesize. In this case it is 3769

 

ls -l afile -rw------- ... 3769 Nov 2 13:41 <filename>

 

dd if=/dev/urandom of=afile bs=3769 count=1 conv=notrunc 

This will write random characters over the entire file. Copy a disk partition to a file on a different partition. Warning!! Do not copy a partition to the same partition.

 

dd if=/dev/sdb2 of=/home/sam/partition.image bs=4096 conv=notrunc,noerror 

This will make a file that is an exact duplicate of the sdb2 partition. You can substitue hdb, sda, hda, or whatever the disk is called. Or

 

dd if=/dev/sdb2 ibs=4096 | gzip > partition.image.gz conv=noerror

Makes a gzipped archive of the entire partition. To restore use:

 

 | gunzip >

for bzip2(slower,smaller), substitute bzip2 and bunzip2, and name the file

 

.bz2

Restore a disk partition from an image file.

 

dd if=/home/sam/partition.image of=/dev/sdb2 bs=4096 conv=notrunc,noerror

This way you can get a large hard drive and partition it so you can back up your root partition. If you mess up your root partition, you just boot from the helix cd and restore the image. To covert a file to all uppercase:[/color]

 

dd if=filename of=filename conv=ucase 

Make a ramdrive: The Linux kernel usually makes a number a ramdisks you can make into ramdrives. You have to populate the drive with zeroes like so:

 

dd if=/dev/zero of=/dev/ram7 bs=1k count=16384

Makes a 16 MB ramdisk full of zeroes.

 

mke2fs -m0 /dev/ram7 4096

puts a file system on the ramdisk turning it into a ramdrive. Watch this puppy smoke.

 

debian:/home/sam # hdparm -t /dev/ram7
/dev/ram7:
Timing buffered disk reads: 16 MB in 0.02 seconds = 913.92 MB/sec

You only to do the timing once because it's cool. Make the drive again because hdparm is a little hard on ramdrives. You can mount the ramdrive with:

 

mkdir /mnt/mem
mount /dev/ram7 /mnt/mem

Now you can use the drive like a hard drive. This is particularly superb for working on large documents or programming. You can copy the large file or programming project to the ramdrive, which on my machine is at least 27 times as fast as /dev/sda, and ever time you save the huge document, or need to do a compile it's like your machine is running on nitromethane. The only thing is the ramdrive is volatile. If you lose power, or lock up the data on the ramdrive is lost. Use a reliable machine during clear skies if you use a ramdrive. Copy ram memory to a file:

 

dd if=/dev/mem of=/home/sam/mem.bin bs=1024 

The device

 

/dev/mem

is your system memory. You can actually copy any block or character device to a file using dd. Memory capture on a fast system, with bs=1024 takes about 60 seconds, a 120 GB HDD about an hour, a CD to hard drive about 10 minutes, a floppy to a hard drive about 2 minutes. With dd, your floppy drive images will not change at all. If you have a bootable DOS diskette, and you save it to your HDD as an image file, when you restore that image to another floppy it will be bootable. dd will print to the terminal window if you omit the

 

of=/dev/output

part.

 

dd if=/home/sam/myfile 

will print the file myfile to the terminal window. To search the system memory: [/color]

 

dd if=/dev/mem | hexdump -C | grep 'some-string-of-words-in-the-file-you-forgot-to-save-before-the-power-failed'

If you need to cover your tracks quickly, put the following commands in a script to overwrite system ram with zeroes. Don't try this for fun.

 

mkdir /mnt/mem

 

mount -t ramfs /dev/mem /mnt/mem

 

dd if=/dev/zero > /mnt/mem/bigfile.file

This will overwrite all unprotected memory structures with zeroes, and freeze the machine so you have to reboot (Caution, this also prevents committment of the file system journal and could trash the file system). If you are just curious about what might be on you disk drive, or what an MBR looks like, or maybe what is at the very end of your disk:

 

dd if=/dev/sda count=1 | hexdump -C 

Will show you sector 1, or the MBR. The bootstrap code and partition table are in the MBR. To see the end of the disk you have to know the total number of sectors for the disk, and the disk has to be set up with Maximum Addressable Sector equal to Maximum Native Address. The helix CD has a utility to set this correctly. In the dd command your seek value will be one less than MNA of the disk. For a 120 GB Seagate SATA drives

 

dd if=/dev/sda of=home/sam/myfile skip=234441646 default bs=512

, So this reads sector for sector, and writes the last sector to myfile. Even with LBA addressing Disks still secretly are read in sectors, cylinders, and heads. There are 63 sectors per cylinder, and 255 heads per cylinder. Then there is a total cylinder count for the disk. You multiply out 512x63x255=bytes per cylinder. 63x255=sectors per cylinder. With 234441647 total sectors, and 16065 sectors per cylinder, you get some trailing sectors which do not make up an entire cylinder, 14593.317584812. This leaves you with 5102 sectors which cannot be partitioned because to be in a partition you have to be a whole cylinder. It's like having part of a person. That doesn't really count as a person. So, what happens to these sectors? They become surplus sectors after the last partition. You can't ordinarily read in there with an operating system. But, dd can. It is really a good idea to check for anything writing to surplus sectors. For our Seagate 120 GB drive you subtract total sectors(234441647)-(5102) which don't make up a whole cylinder=234436545 partitionable sectors.

 

dd if=/dev/sda of=/home/sam/myfile skip=234436545 

This writes the last 5102 sectors to myfile. Launch midnight commander (mc) to view the file. If there is something in there, you do not need it for anything. In this case you would write over it with random characters:

 

dd if=/dev/urandom of=/dev/sda bs=512 seek=234436545 

Will overwrite the 5102 surplus sectors on our 120 GB Seagate drive. Block size: One cylinder in LBA mode = 255 heads * 63 sectors per track = 16065 sectors = 16065 * 512 bytes = 16065b. The b means '* 512'. 32130b represents a two cylinder block size. When using cylinder sized block sizes you never need to worry about that last fraction of a block not being copied because partitions are made of a whole number of cylinders. Partitions cannot contain partial cylinders. One cylinder is 8,225,280 bytes. If you want to check out some random area of the disk:

 

dd if=/dev/sda of=/home/sam/myfile bs=4096 skip=2000 count=1000

Will give you 8,000 sectors in myfile, after the first 16,000 sectors. You can open that file with a hex editor, edit some of it, and write the edited part back to disk:

 

dd if=/home/sam/myfile of=/dev/sda bs=4096 seek=2000 count=1000 

So there you got yourself a disk editor. It's not the best, but it works. You can make a boot floppy:With the

 

boot.img file

, which is pretty easy to get. You just need a program that will literally start writing at sector 1. dd if=boot.img of=/dev/fd0 bs=1440k This makes a bootable disk you can add stuff to. If you want to make a partition image on another machine: on source machine: Boot both machines with the helix CD just to be extra sure. Then,

 

dd if=/dev/hda bs=16065b | netcat targethost-IP 1234 

On target machine:

 

netcat -l -p 1234 | dd of=/dev/hdc bs=16065b

Netcat is a program, available by default, on almost every linux installation. It is like a swiss army knife of networking. In the preceding example netcat and dd are piped to one another. One of the functions of the linux kernel is to make pipes. The pipe character looks like two little lines on top of one another, both vertical. Here is how this command behaves: This byte size is a cylinder. bs=16065b equals one cylinder on an LBA drive. The dd command is piped to netcat, which takes as its arguments the IP address of the target(like 192.168.0.1, or any IP address with an open port) and what port you want to use(1234). Alert!! Don't hit enter yet. Hit enter on the target machine, hit enter on the source machine. This is kind of how Norton Ghost works to image a drive to another machine. Ok, say you want to find out if your girlfriend or wife is cheating on you, having cyber whoopie, or just basically misbehaving with her computer. Even if the computer is secured with a password you can boot with the: www.efense.com/helix CD and search the entire drive partition for text strings:

 

dd if=/dev/sda2 bs=16065 | hexdump -C | grep 'I really don't love him anymore.' 

Will search the whole drive partition for the text string specified between the single quotes. Searching an entire disk partition several times can be quite tedious. This particular command string prints the search results to the screen, with the offset where it is located in the partition. dd works in the decimal system. Disk offsets work in hexidecimal. Say you found that text string in your partition at offset 0x020d0d90h. You convert that to decimal with one of the many calculators found in Linux. This is decimal offset 34409872. Dividing by 512 per sector we get 67206.78125.

 

dd if=/dev/sda2 bs=16065 skip=2140 count=3 | less 

This will print to the screen so you don't accidentally write a file over free disk space, which may hold deleted files you want to search. With this method you search all the deleted files, any chat activity, and emails. It works no matter what security is being employed on the machine. It works with NTFS, ext2, ext3, reiserfs, swap, and FAT partitions. But, it is illegal to use this method on a computer you aren't authorized to search. People can be sued, or imprisoned for unauthorized searches. On a related note, you can write the system memory to a CD. This is useful for documenting memory contents without contaminating the HDD. I recommend using a CD-RW so you can practice a little. This doesn't involve dd, but it's cool.

 

cdrecord dev=ATAPI:0,1,0 -raw tsize=700000000 driveropts=burnfree /dev/mem 

to find the cdwriter: [/color]

 

cdrecord -scanbus=ATAPI 

dd will not copy or erase an HPA or host protected area. If used properly dd will erase a disk completely, but not as well as using the hardware secure erase, security erase unit command

 

 This method records raw, so you have to do a:

 

dd if=/dev/hdd | less

to view the recorded memory. Searching the recorded memory is as above: [/color]

 

dd if=/dev/hdd | hexdump -C | grep 'string' 

string is any ascii sequence, hex sequence (must be separated with a space: '55<space>aa<space>09' searches for the hex string '55aa09'), list:

 

'[[:alnum:]]' any alphanumeric characters 
'[[:alpha:]]' any alpha character
'[[:digit:]]' any numeric character
'[[:blank:]]' tabs and spaces
'[[:lower:]]' any lower case alpha characters
'[[:upper:]]' any uppercase alpha character
'[[:cntrl:]]' ASCII characters 000 thru 037, and 177 octal
'[[:graph:]]' [:alnum:] and [:punct:]
'[[:punct:]]' any punctuation character ` ! ' # $ % ' ( ) * + - . / : ; < = > ? @ [ \ ] ^ _ { | } ~
'[[:space:]]' tab, newline, vertical tab, form feed, carriage return, and space '[[:xdigit:]]' any hex digit ranges('[a-d]' = any, or all abcd, '[0-9]' = any, or all 0123456789)

You can back up your MBR: [/color]

 

dd if=/dev/sda of=mbr.bin count=1 

Put this on a floppy you make with: [/color]

 

dd if=boot.img of=/dev/fd0 

Along with dd. Boot from the floppy and: [/color]

 

dd if=mbr.bin of=/dev/sda count=1 

Will restore the MBR. I back up all my floppies to HDD. Floppies don't last forever, so I do: [/color]

 

dd if=/dev/fd0 of=/home/sam/floppies/backup.bin conv=notrunc 

If my floppy fails, I can make unlimited copies: [/color]

 

dd if=/home/sam/floppies/backup.bin of=/dev/fd0 conv=notrunc 

Here is a command line to read your BIOS, and interfaces: [/color]

 

dd if=/dev/mem bs=1k skip=768 count=256 2>/dev/null | strings -n 8 

There is a variation of dd for rescuing data off defective media, such as a hard drive with some bad sectors. It is called dd_rescue. It is available here: www.garloff.de/kurt/linux/ddrescue/ The Department of Defense implementation of dd is called dcfldd, and has some features like a progress monitor so you can time your trips to get coffee. dcfldd.sourceforge.net Sdd is useful when input block size is different than output block size, and will succeed in some instances dd fails. linux.maruhn.com/sec/sdd.html This is one of the best links on dd I haven't written: www.softpanorama.org/Tools/dd.shtml Note that sending a SIGUSR1 signal to a running 'dd' process makes it print to standard error the number of records read and written so far, then to resume copying.

 

$ dd if=/dev/zero of=/dev/null& pid=$!
$ kill -USR1 $pid; sleep 1; kill $pid

10899206+0 records in 10899206+0 records out

BLOCKS and BYTES may be followed by the following multiplicative suffixes: c 1, w 2, b 512, kB 1000, K 1024, MB 1000*1000, M 1024*1024, GB 1000*1000*1000, G 1024*1024*1024 So,

 

dd if=/dev/sda of=/dev/sdb bs=1GB

Will use one gigabyte block sizes. bs=4b would give dd a block size of 4 disk sectors. 1 sector=512 bytes. bs=4k would indicate dd use a 4 kilobyte block size. I have found bs=4k to be the fastest for copying disk drives on a modern machine. OPERANDS The following operands are supported:

 

if=file

Specifies the input path. Standard input is the default.

 

of=file

Specifies the output path. Standard output is the default. seek=blocks Skip this many blocks in the output file.

 

ibs=n

Specifies the input block size in n bytes (default is 512).

 

obs=n

Specifies the output block size in n bytes (default is 512). If no conversion other than

 

sync, noerror, and, notrunc

is specified, each input block is copied to the output as a single block without aggregating short blocks.

 

cbs=n

Specifies the conversion block size for block and unblock in bytes by n (default is 0). If cbs= is omitted or given a value of 0, using

 

block or unblock

produces unspecified results. This option is used only if ASCII or EBCDIC conversion is specified.

 

ascii and asciib

operands, the input is handled as described for the unblock operand except that characters are converted to ASCII before the trailing SPACE characters are deleted.

 

ebcdic, ebcdicb, ibm, and ibmb

operands, the input is handled as described for the block operand except that the characters are converted to EBCDIC or IBM EBCDIC after the trailing SPACE characters are added.

 

files=n

Copies and concatenates n input files before terminating (makes sense only where input is a magnetic tape or similar device).

 

skip=n

Skips n input blocks (using the specified input block size) before starting to copy. On seekable files, the implementation reads the blocks or seeks past them. On non-seekable files, the blocks are read and the data is discarded.

 

iseek=n

Seeks n blocks from beginning of input file before copying (appropriate for disk files, where skip can be incredibly slow).

 

oseek=n

Seeks n blocks from beginning of output file before copying.

 

seek=n

Skips n blocks (using the specified output block size) from beginning of output file before copying. On non-seekable files, existing blocks are read and space from the current end-of-file to the specified offset, if any, is filled with null bytes. On seekable files, the implementation seeks to the specified offset or reads the blocks as described for non-seekable files.

 

count=n

Copies only n input blocks.

 

conv=value

[,value. . . ] Where values are comma-separated symbols from the following list:

 

conv=notrunc

Tells dd not to abbreviate blocks of all zero value, or multiple adjacent blocks of zeroes, with five asterisks (when you want to maintain size) Do not use notrunc for copying a larger volume to a smaller volume. Without

 

conv=notrunc

Do not truncate the output file.

 

ascii

Converts EBCDIC to ASCII.

 

asciib

Converts EBCDIC to ASCII using BSD-compatible character translations.

 

ebcdic

Converts ASCII to EBCDIC. If converting fixed-length ASCII records without NEWLINEs, sets up a pipeline with

 

dd conv=unblock

beforehand.

 

ebcdicb

Converts ASCII to EBCDIC using BSD-compatible character translations. If converting fixed-length ASCII records without NEWLINEs, sets up a pipeline with

 

dd conv=unblock

beforehand.

 

ibm

Slightly different map of ASCII to EBCDIC. If converting fixed-length ASCII records without NEWLINEs, sets up a pipeline with dd

 

dd conv=unblock

beforehand.

 

ibmb

Slightly different map of ASCII to EBCDIC using BSD-compatible character translations. If converting fixed-length ASCII records without NEWLINEs, sets up a pipeline with

 

dd conv=unblock

beforehand. The

 

ascii (or asciib), ebcdic (or ebcdicb), and ibm (or ibmb)

values are mutually exclusive. block Treats the input as a sequence of NEWLINE-terminated or EOF-terminated variable-length records independent of the input block boundaries. Each record is converted to a record with a fixed length specified by the conversion block size. Any NEWLINE character is removed from the input line. SPACE characters are appended to lines that are shorter than their conversion block size to fill the block. Lines that are longer than the conversion block size are truncated to the largest number of characters that will fit into that size. The number of truncated lines is reported. unblock Converts fixed-length records to variable length. Reads a number of bytes equal to the conversion block size (or the number of bytes remaining in the input, if less than the conversion block size), delete all trailing SPACE characters, and append a NEWLINE character. The block and unblock values are mutually exclusive.

 

lcase

Maps upper-case characters specified by the LC_CTYPE keyword tolower to the corresponding lower-case character. Characters for which no mapping is specified are not modified by this conversion.

 

ucase

Maps lower-case characters specified by the LC_CTYPE keyword toupper to the corresponding upper-case character. Characters for which no mapping is specified are not modified by this conversion. The lcase and ucase symbols are mutually exclusive.

 

swab

Swaps every pair of input bytes. If the current input record is an odd number of bytes, the last byte in the input record is ignored.

 

noerror

Does not stop processing on an input error. When an input error occurs, a diagnostic message is written on standard error, followed by the current input and output block counts in the same format as used at completion. If the

 

sync

conversion is specified, the missing input is replaced with null bytes and processed normally. Otherwise, the input block will be omitted from the output. notrunc Does not truncate the output file. Preserves blocks in the output file not explicitly written by this invocation of dd. (See also the preceding

 

of=file

operand.)

 

sync

Pads every input block to the size of the ibs= buffer, appending null bytes. (If either block or unblock is also specified, appends SPACE characters, rather than null bytes.) ENVIRONMENT VARIABLES The following environment variables affect the messages and errors messages of dd:

 

LANG

Provide a default value for the internationalisation variables that are unset or null. If

 

LANG

is unset or null, the corresponding value from the implementation-dependent default locale will be used. If any of the internationalisation variables contains an invalid setting, the utility will behave as if none of the variables had been defined.

 

LC_ALL

If set to a non-empty string value, override the values of all the other internationalisation variables.

 

LC_CTYPE

Determine the locale for the interpretation of sequences of bytes of text data as characters (for example, single- as opposed to multi-byte characters in arguments and input files), the classification of characters as upper- or lower-case, and the mapping of characters from one case to the other.

 

LC_MESSAGES

Determine the locale that should be used to affect the format and contents of diagnostic messages written to standard error and informative messages written to standard output.

 

NLSPATH

Determine the location of message catalogues for the processing of LC_MESSAGES.

To top

backup con dd e gzip

http://wiki.linuxquestions.org/wiki/Dd

The <tt>dd</tt> command copies data from one place to another. Sometimes cat can do the same thing (with redirection), but dd has options to translate data, selectively copy only part of a data stream, and buffer its reads and writes.

<tt>dd</tt> can copy a CD to an ISO file, copy one partition to another, or restore an image file to a disk. Using the <tt>seek</tt> and <tt>count</tt> options, an individual sector of a disk can be extracted without having to wait for the entire rest of the disk to be read.

<tt>dd</tt> may also be used to extract data from or insert data into arbitrary positions in a file. This can be useful as a way of working with binary files from the command line.

Contents

[hide]

 

 

[edit] Examples

The main options to be concerned about are if= (input file) and of= (output file). By default, <tt>dd</tt> reads from stdin and writes to stdout. Here are some examples of how dd may be used:

[edit] Creating a hard drive backup directly to another hard drive

# dd if=/dev/hda of=/dev/sda conv=noerror,sync bs=4k

This command is used often to create a backup of a drive (/dev/hda) directly to another hard drive (/dev/sda). (The device name /dev/hda is typical of an IDE hard drive, the device /dev/sda is typical of a USB disk.) This works only if the hard drive has enough storage to accommodate the source drive's filesystem. The advantage of this is that you do not have to mount the hard drive to make a backup and the only reference to hda is in /dev and in the command which is usually in a script in cron.

The option "bs=4k" is used to specify the block size used in the copy. The default for the dd command is 512 bytes: use of this small block size can result in significantly slower copying. However, the tradeoff with larger block sizes is that when an error is encountered, the remainder of the block is filled with zero-bytes. So if you increase your block size when copying a failing device, you'll lose more data but also spend less time trying to read broken sectors. Tools like dd_rescue and dd_rhelp can provide a more flexible solution in such cases, combining the speed of a large block size for the regions without errors with finer-grained block-copies for regions with errors.


[edit] Creating a hard drive backup image

# dd if=/dev/hda | gzip > /mnt/hdb1/system_drive_backup.img.gz

Here <tt>dd</tt> is making an image of the first harddrive, and piping it through the gzip compression program. The compressed image is then placed in a file on a seperate drive. To reverse the process:

# gzip -dc /mnt/hdb1/system_drive_backup.img.gz | dd of=/dev/hda

Here, gzip is decompressing (the -d switch) the file, sending the results to stdout (the -c switch), which are piped to dd, and then written to /dev/hda.

[edit] Copy floppy

# dd if=/dev/fd0 of=/tmp/floppy.img bs=10240

That will copy the contents of the floppy to a file. Then, to put the image onto a new floppy, swap "if" and "of" params.

# dd if=/tmp/floppy.img of=/dev/fd0 bs=10240

[edit] Backing up your Master Boot Record (MBR).

You should do this before you edit your partition table so that you can put it back if you mess things up.

# dd if=/dev/hda of=/root/hda.boot.mbr bs=512 count=1

If things mess up, you can boot with Knoppix, mount the partition containing /root (hda1 in this example) and put back the MBR with the command:

# dd if=/mnt/hda1/root/hda.boot.mbr of=/dev/hda bs=512 count=1

Obviously, if you have a GPT system (like the intel mac for instance) this will need some adjustment.

see: forum.onmac.net/showthread.php

You can backup only the MBR and exclude the partition table with the command:

 # dd if=/dev/hda of=/root/hda.mbr.noparttab bs=446 count=1

[edit] Getting around file size limitations using split

When making images, it's quite easy to run up against various file size limitations. One way to work around a given file size limitation is to use the split command.

# dd if=/dev/hda1 | gzip -c | split -b 2000m - /mnt/hdc1/backup.img.gz.
  1. This example is using dd to take an image of the first partition on the first harddrive.
  2. The results are passed through to gzip for compression
    • The -c option switch is used to output the result to stdout.
  3. The compressed image is then piped to the split tool
    • The -b 2000m switch tells split how big to make the individual files. You can use k and m to tell switch kilobytes and megabytes (this option uses bytes by default).
    • The - option tells split to read from stdin. Otherwise, split would interpret the /mnt/hdc1... as the file to be split.
    • The /mnt/hdc1... is the prefix for the created files. Split will create files named backup.img.gz.aa, backup.img.gz.ab, etc.

To restore the multi-file backup, do the following:

# cat /mnt/hdc1/backup.img.gz.* | gzip -dc | dd of=/dev/hda1
  1. Cat recombines contents of the compressed and split image files to stdout, in order.
  2. Results are piped through gzip for decompression.
  3. And are then written to the first partition of the hard drive with dd.

[edit] Creating empty disk images

To create an empty disk image, to be used as the disk for an emulator for example, one can get data from /dev/zero. To create a 10mb image:

$ dd if=/dev/zero of=myimage bs=1024 count=10240

A clever alternative is:

$ dd of=myimage bs=1024 count=0 seek=10240

Here we don't write anything, not even zeroes, we just seek 10mb into the file and close it. The result is a sparse file that is implicitly full of 10mb of zeroes, but that takes no disk space. <tt>ls -l</tt> will report 10mb, while <tt>du</tt> and <tt>df</tt> will report 0. When the file is written to, either as an emulator disk or a loopback device, Linux will allocate disk space for the data. <tt>ls</tt> will still show 10mb, while <tt>du</tt> will gradually approach 10mb.

For swap images, where it's more important to reserve the data than to save disk space, a non-sparse file is better.

[edit] Jargon File Entry

This is what the Jargon File has to say about dd:

<dl><dd>[Unix: from IBM JCL] Equivalent to cat or BLT. Originally the name of a Unix copy command with special options suitable for block-oriented devices; it was often used in heavy-handed system maintenance, as in "Let's dd the root partition onto a tape, then use the boot PROM to load it back on to a new disk". The Unix dd(1)was designed with a weird, distinctly non-Unixy keyword option syntax reminiscent of IBM System/360 JCL (which had an elaborate DD "Dataset Definition" specification for I/O devices); though the command filled a need, the interface design was clearly a prank. The jargon usage is now very rare outside Unix sites and now nearly obsolete even there, as dd(1) has been deprecated for a long time (though it has no exact replacement). The term has been displaced by BLT or simple English "copy". </dd></dl>

Although deprecated, <tt>dd</tt> is still widely in use on many systems.

[edit] See also

[edit] External links

This is the most comprehensive documentation and example sheet for one of the most useful, and least understood linux commands, called "dd". This command has been part of UNIX since the 1970's. It is a bitstream duplicator for copying data, but can use input or output pipes to another command.

<dl><dd>This article is based, in whole or in part, on entry or entries in the Jargon File. </dd></dl>

Retrieved from "http://wiki.linuxquestions.org/wiki/Dd"

 

To top