centos guestfish man page on unix.com

guestfish(1) Virtualization Support guestfish(1)

NAME
guestfish - the guest filesystem shell

SYNOPSIS
guestfish [--options] [commands]

guestfish

guestfish [--ro|--rw] -a disk.img

guestfish [--ro|--rw] -a disk.img -m dev[:mountpoint]

guestfish -d libvirt-domain

guestfish [--ro|--rw] -a disk.img -i

guestfish -d libvirt-domain -i

WARNING
Using guestfish in read/write mode on live virtual machines can be dangerous, potentially causing disk corruption. Use the --ro (read-
only) option to use guestfish safely if the disk image or virtual machine might be live.

DESCRIPTION
Guestfish is a shell and command-line tool for examining and modifying virtual machine filesystems. It uses libguestfs and exposes all of
the functionality of the guestfs API, see guestfs(3).

Guestfish gives you structured access to the libguestfs API, from shell scripts or the command line or interactively. If you want to
rescue a broken virtual machine image, you should look at the virt-rescue(1) command.

EXAMPLES
As an interactive shell
$ guestfish

Welcome to guestfish, the guest filesystem shell for
editing virtual machine filesystems.

Type: 'help' for a list of commands
'man' to read the manual
'quit' to quit the shell

><fs> add-ro disk.img
><fs> run
><fs> list-filesystems
/dev/sda1: ext4
/dev/vg_guest/lv_root: ext4
/dev/vg_guest/lv_swap: swap
><fs> mount /dev/vg_guest/lv_root /
><fs> cat /etc/fstab
# /etc/fstab
# Created by anaconda
[...]
><fs> exit

From shell scripts
Create a new "/etc/motd" file in a guest or disk image:

guestfish <<_EOF_
add disk.img
run
mount /dev/vg_guest/lv_root /
write /etc/motd "Welcome, new users"
_EOF_

List the LVM logical volumes in a disk image:

guestfish -a disk.img --ro <<_EOF_
run
lvs
_EOF_

List all the filesystems in a disk image:

guestfish -a disk.img --ro <<_EOF_
run
list-filesystems
_EOF_

On one command line
Update "/etc/resolv.conf" in a guest:

guestfish
add disk.img : run : mount /dev/vg_guest/lv_root / :
write /etc/resolv.conf "nameserver 1.2.3.4"

Edit "/boot/grub/grub.conf" interactively:

guestfish --rw --add disk.img
--mount /dev/vg_guest/lv_root
--mount /dev/sda1:/boot
edit /boot/grub/grub.conf

Mount disks automatically
Use the -i option to automatically mount the disks from a virtual machine:

guestfish --ro -a disk.img -i cat /etc/group

guestfish --ro -d libvirt-domain -i cat /etc/group

Another way to edit "/boot/grub/grub.conf" interactively is:

guestfish --rw -a disk.img -i edit /boot/grub/grub.conf

As a script interpreter
Create a 100MB disk containing an ext2-formatted partition:

#!/usr/bin/guestfish -f
sparse test1.img 100M
run
part-disk /dev/sda mbr
mkfs ext2 /dev/sda1

Start with a prepared disk
An alternate way to create a 100MB disk called "test1.img" containing a single ext2-formatted partition:

guestfish -N fs

To list what is available do:

guestfish -N help | less

Remote drives
Access a remote disk using ssh:

guestfish -a ssh://example.com/path/to/disk.img

Remote control
eval "`guestfish --listen`"
guestfish --remote add-ro disk.img
guestfish --remote run
guestfish --remote lvs

OPTIONS
--help
Displays general help on options.

-h
--cmd-help
Lists all available guestfish commands.

-h cmd
--cmd-help cmd
Displays detailed help on a single command "cmd".

-a image
--add image
Add a block device or virtual machine image to the shell.

The format of the disk image is auto-detected. To override this and force a particular format use the --format=.. option.

Using this flag is mostly equivalent to using the "add" command, with "readonly:true" if the --ro flag was given, and with "format:..."
if the --format=... flag was given.

-a URI
--add URI
Add a remote disk. See "ADDING REMOTE STORAGE".

-c URI
--connect URI
When used in conjunction with the -d option, this specifies the libvirt URI to use. The default is to use the default libvirt
connection.

--csh
If using the --listen option and a csh-like shell, use this option. See section "REMOTE CONTROL AND CSH" below.

-d libvirt-domain
--domain libvirt-domain
Add disks from the named libvirt domain. If the --ro option is also used, then any libvirt domain can be used. However in write mode,
only libvirt domains which are shut down can be named here.

Domain UUIDs can be used instead of names.

Using this flag is mostly equivalent to using the "add-domain" command, with "readonly:true" if the --ro flag was given, and with
"format:..." if the --format=... flag was given.

-D
--no-dest-paths
Don't tab-complete paths on the guest filesystem. It is useful to be able to hit the tab key to complete paths on the guest
filesystem, but this causes extra "hidden" guestfs calls to be made, so this option is here to allow this feature to be disabled.

--echo-keys
When prompting for keys and passphrases, guestfish normally turns echoing off so you cannot see what you are typing. If you are not
worried about Tempest attacks and there is no one else in the room you can specify this flag to see what you are typing.

-f file
--file file
Read commands from "file". To write pure guestfish scripts, use:

#!/usr/bin/guestfish -f

--format=raw|qcow2|..
--format
The default for the -a option is to auto-detect the format of the disk image. Using this forces the disk format for -a options which
follow on the command line. Using --format with no argument switches back to auto-detection for subsequent -a options.

For example:

guestfish --format=raw -a disk.img

forces raw format (no auto-detection) for "disk.img".

guestfish --format=raw -a disk.img --format -a another.img

forces raw format (no auto-detection) for "disk.img" and reverts to auto-detection for "another.img".

If you have untrusted raw-format guest disk images, you should use this option to specify the disk format. This avoids a possible
security problem with malicious guests (CVE-2010-3851). See also "add".

-i
--inspector
Using virt-inspector(1) code, inspect the disks looking for an operating system and mount filesystems as they would be mounted on the
real virtual machine.

Typical usage is either:

guestfish -d myguest -i

(for an inactive libvirt domain called myguest), or:

guestfish --ro -d myguest -i

(for active domains, readonly), or specify the block device directly:

guestfish --rw -a /dev/Guests/MyGuest -i

Note that the command line syntax changed slightly over older versions of guestfish. You can still use the old syntax:

guestfish [--ro] -i disk.img

guestfish [--ro] -i libvirt-domain

Using this flag is mostly equivalent to using the "inspect-os" command and then using other commands to mount the filesystems that were
found.

--keys-from-stdin
Read key or passphrase parameters from stdin. The default is to try to read passphrases from the user by opening "/dev/tty".

--listen
Fork into the background and listen for remote commands. See section "REMOTE CONTROL GUESTFISH OVER A SOCKET" below.

--live
Connect to a live virtual machine. (Experimental, see "ATTACHING TO RUNNING DAEMONS" in guestfs(3)).

-m dev[:mountpoint[:options[:fstype]]]
--mount dev[:mountpoint[:options[:fstype]]]
Mount the named partition or logical volume on the given mountpoint.

If the mountpoint is omitted, it defaults to "/".

You have to mount something on "/" before most commands will work.

If any -m or --mount options are given, the guest is automatically launched.

If you don't know what filesystems a disk image contains, you can either run guestfish without this option, then list the partitions,
filesystems and LVs available (see "list-partitions", "list-filesystems" and "lvs" commands), or you can use the virt-filesystems(1)
program.

The third (and rarely used) part of the mount parameter is the list of mount options used to mount the underlying filesystem. If this
is not given, then the mount options are either the empty string or "ro" (the latter if the --ro flag is used). By specifying the
mount options, you override this default choice. Probably the only time you would use this is to enable ACLs and/or extended
attributes if the filesystem can support them:

-m /dev/sda1:/:acl,user_xattr

Using this flag is equivalent to using the "mount-options" command.

The fourth part of the parameter is the filesystem driver to use, such as "ext3" or "ntfs". This is rarely needed, but can be useful if
multiple drivers are valid for a filesystem (eg: "ext2" and "ext3"), or if libguestfs misidentifies a filesystem.

--network
Enable QEMU user networking in the guest.

-N type
--new type
-N help
Prepare a fresh disk image formatted as "type". This is an alternative to the -a option: whereas -a adds an existing disk, -N creates
a preformatted disk with a filesystem and adds it. See "PREPARED DISK IMAGES" below.

-n
--no-sync
Disable autosync. This is enabled by default. See the discussion of autosync in the guestfs(3) manpage.

--pipe-error
If writes fail to pipe commands (see "PIPES" below), then the command returns an error.

The default (also for historical reasons) is to ignore such errors so that:

><fs> command_with_lots_of_output | head

doesn't give an error.

--progress-bars
Enable progress bars, even when guestfish is used non-interactively.

Progress bars are enabled by default when guestfish is used as an interactive shell.

--no-progress-bars
Disable progress bars.

--remote[=pid]
Send remote commands to $GUESTFISH_PID or "pid". See section "REMOTE CONTROL GUESTFISH OVER A SOCKET" below.

-r
--ro
This changes the -a, -d and -m options so that disks are added and mounts are done read-only.

The option must always be used if the disk image or virtual machine might be running, and is generally recommended in cases where you
don't need write access to the disk.

Note that prepared disk images created with -N are not affected by this option. Also commands like "add" are not affected - you have
to specify the "readonly:true" option explicitly if you need it.

See also "OPENING DISKS FOR READ AND WRITE" below.

--selinux
Enable SELinux support for the guest. See "SELINUX" in guestfs(3).

-v
--verbose
Enable very verbose messages. This is particularly useful if you find a bug.

-V
--version
Display the guestfish / libguestfs version number and exit.

-w
--rw
This changes the -a, -d and -m options so that disks are added and mounts are done read-write.

See "OPENING DISKS FOR READ AND WRITE" below.

-x Echo each command before executing it.

COMMANDS ON COMMAND LINE
Any additional (non-option) arguments are treated as commands to execute.

Commands to execute should be separated by a colon (":"), where the colon is a separate parameter. Thus:

guestfish cmd [args...] : cmd [args...] : cmd [args...] ...

If there are no additional arguments, then we enter a shell, either an interactive shell with a prompt (if the input is a terminal) or a
non-interactive shell.

In either command line mode or non-interactive shell, the first command that gives an error causes the whole shell to exit. In interactive
mode (with a prompt) if a command fails, you can continue to enter commands.

USING launch (OR run)
As with guestfs(3), you must first configure your guest by adding disks, then launch it, then mount any disks you need, and finally issue
actions/commands. So the general order of the day is:

o add or -a/--add

o launch (aka run)

o mount or -m/--mount

o any other commands

"run" is a synonym for "launch". You must "launch" (or "run") your guest before mounting or performing any other commands.

The only exception is that if any of the -i, -m, --mount, -N or --new options were given then "run" is done automatically, simply because
guestfish can't perform the action you asked for without doing this.

OPENING DISKS FOR READ AND WRITE
The guestfish, guestmount(1) and virt-rescue(1) options --ro and --rw affect whether the other command line options -a, -c, -d, -i and -m
open disk images read-only or for writing.

In libguestfs <= 1.10, guestfish, guestmount and virt-rescue defaulted to opening disk images supplied on the command line for write. To
open a disk image read-only you have to do -a image --ro.

This matters: If you accidentally open a live VM disk image writable then you will cause irreversible disk corruption.

In a future libguestfs we intend to change the default the other way. Disk images will be opened read-only. You will have to either
specify guestfish --rw, guestmount --rw, virt-rescue --rw, or change the configuration file "/etc/libguestfs-tools.conf" in order to get
write access for disk images specified by those other command line options.

This version of guestfish, guestmount and virt-rescue has a --rw option which does nothing (it is already the default). However it is
highly recommended that you use this option to indicate that you need write access, and prepare your scripts for the day when this option
will be required for write access.

Note: This does not affect commands like "add" and "mount", or any other libguestfs program apart from guestfish and guestmount.

QUOTING
You can quote ordinary parameters using either single or double quotes. For example:

add "file with a space.img"

rm '/file name'

rm '/"'

A few commands require a list of strings to be passed. For these, use a whitespace-separated list, enclosed in quotes. Strings containing
whitespace to be passed through must be enclosed in single quotes. A literal single quote must be escaped with a backslash.

vgcreate VG "/dev/sda1 /dev/sdb1"
command "/bin/echo 'foo bar'"
command "/bin/echo 'foo'"

ESCAPE SEQUENCES IN DOUBLE QUOTED ARGUMENTS
In double-quoted arguments (only) use backslash to insert special characters:

"a"
Alert (bell) character.

""
Backspace character.

"f"
Form feed character.

"
"
Newline character.

"
"
Carriage return character.

" "
Horizontal tab character.

"v"
Vertical tab character.

"""
A literal double quote character.

"ooo"
A character with octal value ooo. There must be precisely 3 octal digits (unlike C).

"xhh"
A character with hex value hh. There must be precisely 2 hex digits.

In the current implementation "00" and "x00" cannot be used in strings.

"\"
A literal backslash character.

OPTIONAL ARGUMENTS
Some commands take optional arguments. These arguments appear in this documentation as "[argname:..]". You can use them as in these
examples:

add filename

add filename readonly:true

add filename format:qcow2 readonly:false

Each optional argument can appear at most once. All optional arguments must appear after the required ones.

NUMBERS
This section applies to all commands which can take integers as parameters.

SIZE SUFFIX
When the command takes a parameter measured in bytes, you can use one of the following suffixes to specify kilobytes, megabytes and larger
sizes:

k or K or KiB
The size in kilobytes (multiplied by 1024).

KB The size in SI 1000 byte units.

M or MiB
The size in megabytes (multiplied by 1048576).

MB The size in SI 1000000 byte units.

G or GiB
The size in gigabytes (multiplied by 2**30).

GB The size in SI 10**9 byte units.

T or TiB
The size in terabytes (multiplied by 2**40).

TB The size in SI 10**12 byte units.

P or PiB
The size in petabytes (multiplied by 2**50).

PB The size in SI 10**15 byte units.

E or EiB
The size in exabytes (multiplied by 2**60).

EB The size in SI 10**18 byte units.

Z or ZiB
The size in zettabytes (multiplied by 2**70).

ZB The size in SI 10**21 byte units.

Y or YiB
The size in yottabytes (multiplied by 2**80).

YB The size in SI 10**24 byte units.

For example:

truncate-size /file 1G

would truncate the file to 1 gigabyte.

Be careful because a few commands take sizes in kilobytes or megabytes (eg. the parameter to "memsize" is specified in megabytes already).
Adding a suffix will probably not do what you expect.

OCTAL AND HEXADECIMAL NUMBERS
For specifying the radix (base) use the C convention: 0 to prefix an octal number or "0x" to prefix a hexadecimal number. For example:

1234 decimal number 1234
02322 octal number, equivalent to decimal 1234
0x4d2 hexadecimal number, equivalent to decimal 1234

When using the "chmod" command, you almost always want to specify an octal number for the mode, and you must prefix it with 0 (unlike the
Unix chmod(1) program):

chmod 0777 /public # OK
chmod 777 /public # WRONG! This is mode 777 decimal = 01411 octal.

Commands that return numbers usually print them in decimal, but some commands print numbers in other radices (eg. "umask" prints the mode
in octal, preceded by 0).

WILDCARDS AND GLOBBING
Neither guestfish nor the underlying guestfs API performs wildcard expansion (globbing) by default. So for example the following will not
do what you expect:

rm-rf /home/*

Assuming you don't have a directory called literally "/home/*" then the above command will return an error.

To perform wildcard expansion, use the "glob" command.

glob rm-rf /home/*

runs "rm-rf" on each path that matches (ie. potentially running the command many times), equivalent to:

rm-rf /home/jim
rm-rf /home/joe
rm-rf /home/mary

"glob" only works on simple guest paths and not on device names.

If you have several parameters, each containing a wildcard, then glob will perform a Cartesian product.

COMMENTS
Any line which starts with a # character is treated as a comment and ignored. The # can optionally be preceded by whitespace, but not by a
command. For example:

# this is a comment
# this is a comment
foo # NOT a comment

Blank lines are also ignored.

RUNNING COMMANDS LOCALLY
Any line which starts with a ! character is treated as a command sent to the local shell ("/bin/sh" or whatever system(3) uses). For
example:

!mkdir local
tgz-out /remote local/remote-data.tar.gz

will create a directory "local" on the host, and then export the contents of "/remote" on the mounted filesystem to
"local/remote-data.tar.gz". (See "tgz-out").

To change the local directory, use the "lcd" command. "!cd" will have no effect, due to the way that subprocesses work in Unix.

LOCAL COMMANDS WITH INLINE EXECUTION
If a line starts with <! then the shell command is executed (as for !), but subsequently any output (stdout) of the shell command is parsed
and executed as guestfish commands.

Thus you can use shell script to construct arbitrary guestfish commands which are then parsed by guestfish.

For example it is tedious to create a sequence of files (eg. "/foo.1" through "/foo.100") using guestfish commands alone. However this is
simple if we use a shell script to create the guestfish commands for us:

<! for n in `seq 1 100`; do echo write /foo.$n $n; done

or with names like "/foo.001":

<! for n in `seq 1 100`; do printf "write /foo.%03d %d
" $n $n; done

When using guestfish interactively it can be helpful to just run the shell script first (ie. remove the initial "<" character so it is just
an ordinary ! local command), see what guestfish commands it would run, and when you are happy with those prepend the "<" character to run
the guestfish commands for real.

PIPES
Use "command <space> | command" to pipe the output of the first command (a guestfish command) to the second command (any host command).
For example:

cat /etc/passwd | awk -F: '$3 == 0 { print }'

(where "cat" is the guestfish cat command, but "awk" is the host awk program). The above command would list all accounts in the guest
filesystem which have UID 0, ie. root accounts including backdoors. Other examples:

hexdump /bin/ls | head
list-devices | tail -1
tgz-out / - | tar ztf -

The space before the pipe symbol is required, any space after the pipe symbol is optional. Everything after the pipe symbol is just passed
straight to the host shell, so it can contain redirections, globs and anything else that makes sense on the host side.

To use a literal argument which begins with a pipe symbol, you have to quote it, eg:

echo "|"

HOME DIRECTORIES
If a parameter starts with the character "~" then the tilde may be expanded as a home directory path (either "~" for the current user's
home directory, or "~user" for another user).

Note that home directory expansion happens for users known on the host, not in the guest filesystem.

To use a literal argument which begins with a tilde, you have to quote it, eg:

echo "~"

ENCRYPTED DISKS
Libguestfs has some support for Linux guests encrypted according to the Linux Unified Key Setup (LUKS) standard, which includes nearly all
whole disk encryption systems used by modern Linux guests. Currently only LVM-on-LUKS is supported.

Identify encrypted block devices and partitions using "vfs-type":

><fs> vfs-type /dev/sda2
crypto_LUKS

Then open those devices using "luks-open". This creates a device-mapper device called "/dev/mapper/luksdev".

><fs> luks-open /dev/sda2 luksdev
Enter key or passphrase ("key"): <enter the passphrase>

Finally you have to tell LVM to scan for volume groups on the newly created mapper device:

vgscan
vg-activate-all true

The logical volume(s) can now be mounted in the usual way.

Before closing a LUKS device you must unmount any logical volumes on it and deactivate the volume groups by calling "vg-activate false VG"
on each one. Then you can close the mapper device:

vg-activate false /dev/VG
luks-close /dev/mapper/luksdev

WINDOWS PATHS
If a path is prefixed with "win:" then you can use Windows-style drive letters and paths (with some limitations). The following commands
are equivalent:

file /WINDOWS/system32/config/system.LOG

file win:windowssystem32configsystem.log

file WIN:C:WindowsSYSTEM32CONFIGSYSTEM.LOG

The parameter is rewritten "behind the scenes" by looking up the position where the drive is mounted, prepending that to the path, changing
all backslash characters to forward slash, then resolving the result using "case-sensitive-path". For example if the E: drive was mounted
on "/e" then the parameter might be rewritten like this:

win:e:fooar => /e/FOO/bar

This only works in argument positions that expect a path.

UPLOADING AND DOWNLOADING FILES
For commands such as "upload", "download", "tar-in", "tar-out" and others which upload from or download to a local file, you can use the
special filename "-" to mean "from stdin" or "to stdout". For example:

upload - /foo

reads stdin and creates from that a file "/foo" in the disk image, and:

tar-out /etc - | tar tf -

writes the tarball to stdout and then pipes that into the external "tar" command (see "PIPES").

When using "-" to read from stdin, the input is read up to the end of stdin. You can also use a special "heredoc"-like syntax to read up
to some arbitrary end marker:

upload -<<END /foo
input line 1
input line 2
input line 3
END

Any string of characters can be used instead of "END". The end marker must appear on a line of its own, without any preceding or following
characters (not even spaces).

Note that the "-<<" syntax only applies to parameters used to upload local files (so-called "FileIn" parameters in the generator).

EXIT ON ERROR BEHAVIOUR
By default, guestfish will ignore any errors when in interactive mode (ie. taking commands from a human over a tty), and will exit on the
first error in non-interactive mode (scripts, commands given on the command line).

If you prefix a command with a - character, then that command will not cause guestfish to exit, even if that (one) command returns an
error.

REMOTE CONTROL GUESTFISH OVER A SOCKET
Guestfish can be remote-controlled over a socket. This is useful particularly in shell scripts where you want to make several different
changes to a filesystem, but you don't want the overhead of starting up a guestfish process each time.

Start a guestfish server process using:

eval "`guestfish --listen`"

and then send it commands by doing:

guestfish --remote cmd [...]

To cause the server to exit, send it the exit command:

guestfish --remote exit

Note that the server will normally exit if there is an error in a command. You can change this in the usual way. See section "EXIT ON
ERROR BEHAVIOUR".

CONTROLLING MULTIPLE GUESTFISH PROCESSES
The "eval" statement sets the environment variable $GUESTFISH_PID, which is how the --remote option knows where to send the commands. You
can have several guestfish listener processes running using:

eval "`guestfish --listen`"
pid1=$GUESTFISH_PID
eval "`guestfish --listen`"
pid2=$GUESTFISH_PID
...
guestfish --remote=$pid1 cmd
guestfish --remote=$pid2 cmd

REMOTE CONTROL AND CSH
When using csh-like shells (csh, tcsh etc) you have to add the --csh option:

eval "`guestfish --listen --csh`"

REMOTE CONTROL DETAILS
Remote control happens over a Unix domain socket called "/tmp/.guestfish-$UID/socket-$PID", where $UID is the effective user ID of the
process, and $PID is the process ID of the server.

Guestfish client and server versions must match exactly.

Older versions of guestfish were vulnerable to CVE-2013-4419 (see "CVE-2013-4419" in guestfs(3)). This is fixed in the current version.

USING REMOTE CONTROL ROBUSTLY FROM SHELL SCRIPTS
From Bash, you can use the following code which creates a guestfish instance, correctly quotes the command line, handles failure to start,
and cleans up guestfish when the script exits:

#!/bin/bash -

set -e

guestfish[0]="guestfish"
guestfish[1]="--listen"
guestfish[2]="--ro"
guestfish[3]="-a"
guestfish[4]="disk.img"

GUESTFISH_PID=
eval $("${guestfish[@]}")
if [ -z "$GUESTFISH_PID" ]; then
echo "error: guestfish didn't start up, see error messages above"
exit 1
fi

cleanup_guestfish ()
{
guestfish --remote -- exit >/dev/null 2>&1 ||:
}
trap cleanup_guestfish EXIT ERR

guestfish --remote -- run

# ...

REMOTE CONTROL DOES NOT WORK WITH -a ETC. OPTIONS
Options such as -a, --add, -N, --new etc don't interact properly with remote support. They are processed locally, and not sent through to
the remote guestfish. In particular this won't do what you expect:

guestfish --remote --add disk.img

Don't use these options. Use the equivalent commands instead, eg:

guestfish --remote add-drive disk.img

or:

guestfish --remote
><fs> add disk.img

REMOTE CONTROL RUN COMMAND HANGING
Using the "run" (or "launch") command remotely in a command substitution context hangs, ie. don't do (note the backquotes):

a=`guestfish --remote run`

Since the "run" command produces no output on stdout, this is not useful anyway. For further information see
https://bugzilla.redhat.com/show_bug.cgi?id=592910.

PREPARED DISK IMAGES
Use the -N type or --new type parameter to select one of a set of preformatted disk images that guestfish can make for you to save typing.
This is particularly useful for testing purposes. This option is used instead of the -a option, and like -a can appear multiple times (and
can be mixed with -a).

The new disk is called "test1.img" for the first -N, "test2.img" for the second and so on. Existing files in the current directory are
overwritten.

The type briefly describes how the disk should be sized, partitioned, how filesystem(s) should be created, and how content should be added.
Optionally the type can be followed by extra parameters, separated by ":" (colon) characters. For example, -N fs creates a default 100MB,
sparsely-allocated disk, containing a single partition, with the partition formatted as ext2. -N fs:ext4:1G is the same, but for an ext4
filesystem on a 1GB disk instead.

Note that the prepared filesystem is not mounted. You would usually have to use the "mount /dev/sda1 /" command or add the -m /dev/sda1
option.

If any -N or --new options are given, the libguestfs appliance is automatically launched.

EXAMPLES
Create a 100MB disk with an ext4-formatted partition:

guestfish -N fs:ext4

Create a 32MB disk with a VFAT-formatted partition, and mount it:

guestfish -N fs:vfat:32M -m /dev/sda1

Create a blank 200MB disk:

guestfish -N disk:200M

-N disk - create a blank disk
"guestfish -N disk[:size]"

Create a blank disk, size 100MB (by default).

The default size can be changed by supplying an optional parameter.

The optional parameters are:

Name Default value
size 100M the size of the disk image

-N part - create a partitioned disk
"guestfish -N part[:size[:partition]]"

Create a disk with a single partition. By default the size of the disk is 100MB (the available space in the partition will be a tiny bit
smaller) and the partition table will be MBR (old DOS-style).

These defaults can be changed by supplying optional parameters.

The optional parameters are:

Name Default value
size 100M the size of the disk image
partition mbr partition table type

-N fs - create a filesystem
"guestfish -N fs[:filesystem[:size[:partition]]]"

Create a disk with a single partition, with the partition containing an empty filesystem. This defaults to creating a 100MB disk (the
available space in the filesystem will be a tiny bit smaller) with an MBR (old DOS-style) partition table and an ext2 filesystem.

These defaults can be changed by supplying optional parameters.

The optional parameters are:

Name Default value
filesystem ext2 the type of filesystem to use
size 100M the size of the disk image
partition mbr partition table type

-N lv - create a disk with logical volume
"guestfish -N lv[:name[:size[:partition]]]"

Create a disk with a single partition, set up the partition as an LVM2 physical volume, and place a volume group and logical volume on
there. This defaults to creating a 100MB disk with the VG and LV called "/dev/VG/LV". You can change the name of the VG and LV by
supplying an alternate name as the first optional parameter.

Note this does not create a filesystem. Use 'lvfs' to do that.

The optional parameters are:

Name Default value
name /dev/VG/LV the name of the VG and LV to use
size 100M the size of the disk image
partition mbr partition table type

-N lvfs - create a disk with logical volume and filesystem
"guestfish -N lvfs[:name[:filesystem[:size[:partition]]]]"

Create a disk with a single partition, set up the partition as an LVM2 physical volume, and place a volume group and logical volume on
there. Then format the LV with a filesystem. This defaults to creating a 100MB disk with the VG and LV called "/dev/VG/LV", with an ext2
filesystem.

The optional parameters are:

Name Default value
name /dev/VG/LV the name of the VG and LV to use
filesystem ext2 the type of filesystem to use
size 100M the size of the disk image
partition mbr partition table type

-N bootroot - create a boot and root filesystem
"guestfish -N bootroot[:bootfs[:rootfs[:size[:bootsize[:partition]]]]]"

Create a disk with two partitions, for boot and root filesystem. Format the two filesystems independently. There are several optional
parameters which control the exact layout and filesystem types.

The optional parameters are:

Name Default value
bootfs ext2 the type of filesystem to use for boot
rootfs ext2 the type of filesystem to use for root
size 100M the size of the disk image
bootsize 32M the size of the boot filesystem
partition mbr partition table type

-N bootrootlv - create a boot and root filesystem using LVM
"guestfish -N bootrootlv[:name[:bootfs[:rootfs[:size[:bootsize[:partition]]]]]]"

This is the same as "bootroot" but the root filesystem (only) is placed on a logical volume, named by default "/dev/VG/LV". There are
several optional parameters which control the exact layout.

The optional parameters are:

Name Default value
name /dev/VG/LV the name of the VG and LV for root
bootfs ext2 the type of filesystem to use for boot
rootfs ext2 the type of filesystem to use for root
size 100M the size of the disk image
bootsize 32M the size of the boot filesystem
partition mbr partition table type

ADDING REMOTE STORAGE
For API-level documentation on this topic, see "guestfs_add_drive_opts" in guestfs(3) and "REMOTE STORAGE" in guestfs(3).

On the command line, you can use the -a option to add network block devices using a URI-style format, for example:

guestfish -a ssh://root@example.com/disk.img

URIs cannot be used with the "add" command. The equivalent command using the API directly is:

><fs> add /disk.img protocol:ssh server:tcp:example.com username:root

The possible -a URI formats are described below.

-a disk.img
-a file:///path/to/disk.img
Add the local disk image (or device) called "disk.img".

-a nbd://example.com[:port]
-a nbd://example.com[:port]/exportname
-a nbd://?socket=/socket
-a nbd:///exportname?socket=/socket
Add a disk located on Network Block Device (nbd) storage.

The /exportname part of the URI specifies an NBD export name, but is usually left empty.

The optional ?socket parameter can be used to specify a Unix domain socket that we talk to the NBD server over. Note that you cannot mix
server name (ie. TCP/IP) and socket path.

The equivalent API command would be (no export name):

><fs> add "" protocol:nbd server:[tcp:example.com|unix:/socket]

PROGRESS BARS
Some (not all) long-running commands send progress notification messages as they are running. Guestfish turns these messages into progress
bars.

When a command that supports progress bars takes longer than two seconds to run, and if progress bars are enabled, then you will see one
appearing below the command:

><fs> copy-size /large-file /another-file 2048M
/ 10% [#####-----------------------------------------] 00:30

The spinner on the left hand side moves round once for every progress notification received from the backend. This is a (reasonably)
golden assurance that the command is "doing something" even if the progress bar is not moving, because the command is able to send the
progress notifications. When the bar reaches 100% and the command finishes, the spinner disappears.

Progress bars are enabled by default when guestfish is used interactively. You can enable them even for non-interactive modes using
--progress-bars, and you can disable them completely using --no-progress-bars.

PROMPT
You can change or add colours to the default prompt ("><fs>") by setting the "GUESTFISH_PS1" environment variable. A simple prompt can be
set by setting this to an alternate string:

$ GUESTFISH_PS1='(type a command) '
$ export GUESTFISH_PS1
$ guestfish
[...]
(type a command)

You can also use special escape sequences, as described in the table below:

\ A literal backslash character.

[
] Place non-printing characters (eg. terminal control codes for colours) between "[...]". What this does it to tell the readline(3)
library that it should treat this subsequence as zero-width, so that command-line redisplay, editing etc works.

a A bell character.

e An ASCII ESC (escape) character.

A newline.

A carriage return.

NNN
The ASCII character whose code is the octal value NNN.

xNN
The ASCII character whose code is the hex value NN.

EXAMPLES OF PROMPTS
Note these these require a terminal that supports ANSI escape codes.

GUESTFISH_PS1='[e[1;30m]><fs>[e[0;30m] '

A bold black version of the ordinary prompt.

WINDOWS 8
Windows 8 "fast startup" can prevent guestfish from mounting NTFS partitions. See "WINDOWS HIBERNATION AND WINDOWS 8 FAST STARTUP" in
guestfs(3).

GUESTFISH COMMANDS
The commands in this section are guestfish convenience commands, in other words, they are not part of the guestfs(3) API.

help
help
help cmd

Without any parameter, this provides general help.

With a "cmd" parameter, this displays detailed help for that command.

exit
quit
This exits guestfish. You can also use "^D" key.

alloc
allocate
alloc filename size

This creates an empty (zeroed) file of the given size, and then adds so it can be further examined.

For more advanced image creation, see qemu-img(1) utility.

Size can be specified using standard suffixes, eg. "1M".

To create a sparse file, use "sparse" instead. To create a prepared disk image, see "PREPARED DISK IMAGES".

copy-in
copy-in local [local ...] /remotedir

"copy-in" copies local files or directories recursively into the disk image, placing them in the directory called "/remotedir" (which must
exist). This guestfish meta-command turns into a sequence of "tar-in" and other commands as necessary.

Multiple local files and directories can be specified, but the last parameter must always be a remote directory. Wildcards cannot be used.

copy-out
copy-out remote [remote ...] localdir

"copy-out" copies remote files or directories recursively out of the disk image, placing them on the host disk in a local directory called
"localdir" (which must exist). This guestfish meta-command turns into a sequence of "download", "tar-out" and other commands as necessary.

Multiple remote files and directories can be specified, but the last parameter must always be a local directory. To download to the
current directory, use "." as in:

copy-out /home .

Wildcards cannot be used in the ordinary command, but you can use them with the help of "glob" like this:

glob copy-out /home/* .

delete-event
delete-event name

Delete the event handler which was previously registered as "name". If multiple event handlers were registered with the same name, they
are all deleted.

See also the guestfish commands "event" and "list-events".

display
display filename

Use "display" (a graphical display program) to display an image file. It downloads the file, and runs "display" on it.

To use an alternative program, set the "GUESTFISH_DISPLAY_IMAGE" environment variable. For example to use the GNOME display program:

export GUESTFISH_DISPLAY_IMAGE=eog

See also "stat".

This command has one or more optional arguments. See "OPTIONAL ARGUMENTS".

is-chardev
is-chardev-opts
is-chardev path [followsymlinks:true|false]

This returns "true" if and only if there is a character device with the given "path" name.

If the optional flag "followsymlinks" is true, then a symlink (or chain of symlinks) that ends with a chardev also causes the function to
return true.

See also "stat".

This command has one or more optional arguments. See "OPTIONAL ARGUMENTS".

is-config
is-config

This returns true iff this handle is being configured (in the "CONFIG" state).

For more information on states, see guestfs(3).

is-dir
is-dir-opts
is-dir path [followsymlinks:true|false]

This returns "true" if and only if there is a directory with the given "path" name. Note that it returns false for other objects like
files.

If the optional flag "followsymlinks" is true, then a symlink (or chain of symlinks) that ends with a directory also causes the function to
return true.

See also "stat".

This command has one or more optional arguments. See "OPTIONAL ARGUMENTS".

is-fifo
is-fifo-opts
is-fifo path [followsymlinks:true|false]

This returns "true" if and only if there is a FIFO (named pipe) with the given "path" name.

If the optional flag "followsymlinks" is true, then a symlink (or chain of symlinks) that ends with a FIFO also causes the function to
return true.

See also "stat".

This command has one or more optional arguments. See "OPTIONAL ARGUMENTS".

is-file
is-file-opts
is-file path [followsymlinks:true|false]

This returns "true" if and only if there is a regular file with the given "path" name. Note that it returns false for other objects like
directories.

If the optional flag "followsymlinks" is true, then a symlink (or chain of symlinks) that ends with a file also causes the function to
return true.

See also "stat".

This command has one or more optional arguments. See "OPTIONAL ARGUMENTS".

is-lv
is-lv device

This command tests whether "device" is a logical volume, and returns true iff this is the case.

is-socket
is-socket-opts
is-socket path [followsymlinks:true|false]

This returns "true" if and only if there is a Unix domain socket with the given "path" name.

If the optional flag "followsymlinks" is true, then a symlink (or chain of symlinks) that ends with a socket also causes the function to
return true.

See also "stat".

This command has one or more optional arguments. See "OPTIONAL ARGUMENTS".

is-symlink
is-symlink path

This returns "true" if and only if there is a symbolic link with the given "path" name.

See also "stat".

is-whole-device
is-whole-device device

This returns "true" if and only if "device" refers to a whole block device. That is, not a partition or a logical device.

is-zero
is-zero path

This returns true iff the file exists and the file is empty or it contains all zero bytes.

is-zero-device
is-zero-device device

This returns true iff the device exists and contains all zero bytes.

Note that for large devices this can take a long time to run.

isoinfo
isoinfo isofile

This is the same as "isoinfo-device" except that it works for an ISO file located inside some other mounted filesystem. Note that in the
common case where you have added an ISO file as a libguestfs device, you would not call this. Instead you would call "isoinfo-device".

isoinfo-device
isoinfo-device device

"device" is an ISO device. This returns a struct of information read from the primary volume descriptor (the ISO equivalent of the
superblock) of the device.

Usually it is more efficient to use the isoinfo(1) command with the -d option on the host to analyze ISO files, instead of going through
libguestfs.

For information on the primary volume descriptor fields, see http://wiki.osdev.org/ISO_9660#The_Primary_Volume_Descriptor

kill-subprocess
kill-subprocess

This kills the qemu subprocess.

Do not call this. See: "shutdown" instead.

This function is deprecated. In new code, use the "shutdown" call instead.

Deprecated functions will not be removed from the API, but the fact that they are deprecated indicates that there are problems with correct
use of these functions.

launch
run
launch

Internally libguestfs is implemented by running a virtual machine using qemu(1).

You should call this after configuring the handle (eg. adding drives) but before performing any actions.

Do not call "launch" twice on the same handle. Although it will not give an error (for historical reasons), the precise behaviour when you
do this is not well defined. Handles are very cheap to create, so create a new one for each launch.

lchown
lchown owner group path

Change the file owner to "owner" and group to "group". This is like "chown" but if "path" is a symlink then the link itself is changed,
not the target.

Only numeric uid and gid are supported. If you want to use names, you will need to locate and parse the password file yourself (Augeas
support makes this relatively easy).

ldmtool-create-all
ldmtool-create-all

This function scans all block devices looking for Windows dynamic disk volumes and partitions, and creates devices for any that were found.

Call "list-ldm-volumes" and "list-ldm-partitions" to return all devices.

Note that you don't normally need to call this explicitly, since it is done automatically at "launch" time. However you might want to call
this function if you have hotplugged disks or have just created a Windows dynamic disk.

ldmtool-diskgroup-disks
ldmtool-diskgroup-disks diskgroup

Return the disks in a Windows dynamic disk group. The "diskgroup" parameter should be the GUID of a disk group, one element from the list
returned by "ldmtool-scan".

ldmtool-diskgroup-name
ldmtool-diskgroup-name diskgroup

Return the name of a Windows dynamic disk group. The "diskgroup" parameter should be the GUID of a disk group, one element from the list
returned by "ldmtool-scan".

ldmtool-diskgroup-volumes
ldmtool-diskgroup-volumes diskgroup

Return the volumes in a Windows dynamic disk group. The "diskgroup" parameter should be the GUID of a disk group, one element from the
list returned by "ldmtool-scan".

ldmtool-remove-all
ldmtool-remove-all

This is essentially the opposite of "ldmtool-create-all". It removes the device mapper mappings for all Windows dynamic disk volumes

ldmtool-scan
ldmtool-scan

This function scans for Windows dynamic disks. It returns a list of identifiers (GUIDs) for all disk groups that were found. These
identifiers can be passed to other "ldmtool-*" functions.

This function scans all block devices. To scan a subset of block devices, call "ldmtool-scan-devices" instead.

ldmtool-scan-devices
ldmtool-scan-devices 'devices ...'

This function scans for Windows dynamic disks. It returns a list of identifiers (GUIDs) for all disk groups that were found. These
identifiers can be passed to other "ldmtool-*" functions.

The parameter "devices" is a list of block devices which are scanned. If this list is empty, all block devices are scanned.

ldmtool-volume-hint
ldmtool-volume-hint diskgroup volume

Return the hint field of the volume named "volume" in the disk group with GUID "diskgroup". This may not be defined, in which case the
empty string is returned. The hint field is often, though not always, the name of a Windows drive, eg. "E:".

ldmtool-volume-partitions
ldmtool-volume-partitions diskgroup volume

Return the list of partitions in the volume named "volume" in the disk group with GUID "diskgroup".

ldmtool-volume-type
ldmtool-volume-type diskgroup volume

Return the type of the volume named "volume" in the disk group with GUID "diskgroup".

Possible volume types that can be returned here include: "simple", "spanned", "striped", "mirrored", "raid5". Other types may also be
returned.

lgetxattr
lgetxattr path name

Get a single extended attribute from file "path" named "name". If "path" is a symlink, then this call returns an extended attribute from
the symlink.

Extended attribute values are blobs of binary data. If there is no extended attribute named "name", this returns an error.

See also: "lgetxattrs", "getxattr", attr(5).

lgetxattrs
lgetxattrs path

This is the same as "getxattrs", but if "path" is a symbolic link, then it returns the extended attributes of the link itself.

list-devices
list-devices

List all the block devices.

The full block device names are returned, eg. "/dev/sda".

centos man page for guestfish