Linux Manpages Online - man.cx manual pages

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 write mode on live virtual machines, or concurrently with other disk editing tools, can be dangerous, potentially causing disk corruption. The virtual machine must be shut down before you use this command, and disk images must not be edited concurrently.

Use the --ro (read-only) option to use "guestfish" safely if the disk image or virtual machine might be live. You may see strange or inconsistent results if running concurrently with other changes, but with this option you won’t risk disk corruption.

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
Create a 1G 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".

--blocksize=512
--blocksize=4096
--blocksize

This parameter sets the sector size of the disk image. It affects all explicitly added subsequent disks after this parameter. Using --blocksize with no argument switches the disk sector size to the default value which is usually 512 bytes. See also "guestfs_add_drive_opts" in guestfs(3).

-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.

--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.

--key SELECTOR

Specify a key for LUKS, to automatically open a LUKS device when using the inspection.
--key NAME:key:KEY_STRING
--key UUID:key:KEY_STRING
--key all:key:KEY_STRING

"NAME" is the libguestfs device name (eg. "/dev/sda1"). "UUID" is the device UUID. "all" means try the key against any encrypted device.

Use the specified "KEY_STRING" as passphrase.

--key NAME:file:FILENAME
--key UUID:file:FILENAME
--key all:file:FILENAME

Read the passphrase from FILENAME.

--key NAME:clevis
--key UUID:clevis
--key all:clevis

Attempt passphrase-less unlocking for the device with Clevis, over the network. Please refer to "ENCRYPTED DISKS" in guestfs(3) for more information on network-bound disk encryption (NBDE).

Note that if any such option is present on the command line, QEMU user networking will be automatically enabled for the libguestfs appliance.

--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.

If there are multiple encrypted devices then you may need to supply multiple keys on stdin, one per line.

--listen

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

-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 [FILENAME=]TYPE
--new [FILENAME=]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.

--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.

--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
--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

This option is provided for backwards compatibility and does nothing.

-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.

Note that arguments of the commands will be considered as guestfish options if they start with a dash ("-"): you can always separate the guestfish options and the rest of the commands (with their arguments) using a double dash ("--"). For example:

guestfish -- disk_create overlay.qcow2 qcow2 -1 backingfile:image.img

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:

	•		add or -a/--add
	•		launch (aka run)
	•		mount or -m/--mount
	•		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 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.

"\b"

Backspace character.

"\f"

Form feed character.

"\n"

Newline character.

"\r"

Carriage return character.

"\t"

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 "\000" 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).

The size in SI 1000 byte units.

M or MiB

The size in megabytes (multiplied by 1048576).

The size in SI 1000000 byte units.

G or GiB

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

The size in SI 10**9 byte units.

T or TiB

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

The size in SI 10**12 byte units.

P or PiB

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

The size in SI 10**15 byte units.

E or EiB

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

The size in SI 10**18 byte units.

Z or ZiB

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

The size in SI 10**21 byte units.

Y or YiB

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

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 $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, and Windows BitLocker.

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

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

or:

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

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

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

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

vgscan
vg-activate-all true

The filesystems or logical volumes can now be mounted in the usual way.

Before closing an encrypted 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
cryptsetup-close /dev/mapper/name

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:\windows\system32\config\system.log
file WIN:C:\Windows\SYSTEM32\CONFIG\SYSTEM.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:\foo\bar => /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 [filename=]type or --new [filename=]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. You can use a different filename by specifying "filename=" before the type (see examples below).

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 1G, sparsely-allocated disk, containing a single partition, with the partition formatted as ext2. -N fs:ext4:2G is the same, but for an ext4 filesystem on a 2GB 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 1G disk with an ext4-formatted partition, called test1.img in the current directory:

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

Create a blank 200MB disk called blankdisk.img (instead of test1.img):

guestfish -N blankdisk.img=disk:200M

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

Create a blank disk, size 1G (by default).

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

The optional parameters are:

Name Default value
size 1G the size of the disk image

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

Create a disk with a single partition. By default the size of the disk is 1G (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 1G the size of the disk image
partition mbr partition table type

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

Create a disk with a single partition, with the partition containing an empty filesystem. This defaults to creating a 1G 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 1G the size of the disk image
partition mbr partition table type

-N lv - create a disk with logical volume
"guestfish -N [filename=]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 1G 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 1G the size of the disk image
partition mbr partition table type

-N lvfs - create a disk with logical volume and filesystem
"guestfish -N [filename=]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 1G 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 1G the size of the disk image
partition mbr partition table type

-N bootroot - create a boot and root filesystem
"guestfish -N [filename=]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 1G the size of the disk image
bootsize 128M the size of the boot filesystem
partition mbr partition table type

-N bootrootlv - create a boot and root filesystem using LVM
"guestfish -N [filename=]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 1G the size of the disk image
bootsize 128M 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 [AT] 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 ftp://[user@]example.com[:port]/disk.img
-a ftps://[user@]example.com[:port]/disk.img
-a http://[user@]example.com[:port]/disk.img
-a https://[user@]example.com[:port]/disk.img
-a tftp://[user@]example.com[:port]/disk.img
Add a disk located on a remote FTP, HTTP or TFTP server.

The equivalent API command would be:

><fs> add /disk.img protocol:(ftp|...) server:tcp:example.com

-a gluster://example.com[:port]/volname/image
Add a disk image located on GlusterFS storage.

The server is the one running "glusterd", and may be "localhost".

The equivalent API command would be:

><fs> add volname/image protocol:gluster server:tcp:example.com

-a iscsi://example.com[:port]/target-iqn-name[/lun]
Add a disk located on an iSCSI server.

The equivalent API command would be:

><fs> add target-iqn-name/lun protocol:iscsi server:tcp:example.com

-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]

-a rbd:///pool/disk
-a rbd://example.com[:port]/pool/disk
Add a disk image located on a Ceph (RBD/librbd) storage volume.

Although libguestfs and Ceph supports multiple servers, only a single server can be specified when using this URI syntax.

The equivalent API command would be:

><fs> add pool/disk protocol:rbd server:tcp:example.com:port

-a sheepdog://[example.com[:port]]/volume/image
Add a disk image located on a Sheepdog volume.

The server name is optional. Although libguestfs and Sheepdog supports multiple servers, only at most one server can be specified when using this URI syntax.

The equivalent API command would be:

><fs> add volume protocol:sheepdog [server:tcp:example.com]

-a ssh://[user@]example.com[:port]/disk.img
Add a disk image located on a remote server, accessed using the Secure Shell (ssh) SFTP protocol. SFTP is supported out of the box by all major SSH servers.

The equivalent API command would be:

><fs> add /disk protocol:ssh server:tcp:example.com [username:user]

Note that the URIs follow the syntax of RFC 3986: in particular, there are restrictions on the allowed characters for the various components of the URI. Characters such as ":", "@", and "/" must be percent-encoded:

$ guestfish -a ssh://user:pass%40word [AT] example.com/disk.img

In this case, the password is "pass@word".

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 second string ("GUESTFISH_OUTPUT") is printed after the command has been entered and before the output, allowing you to control the colour of the output. A third string ("GUESTFISH_INIT") is printed before the welcome message, allowing you to control the colour of that message. A fourth string ("GUESTFISH_RESTORE") is printed before guestfish exits.

A simple prompt can be set by setting "GUESTFISH_PS1" 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.
	\[
	\]		(These should only be used in "GUESTFISH_PS1".)

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.
	\n		A newline.
	\r		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 that these examples 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.

•

GUESTFISH_PS1='\[\e[1;32m\]><fs>\[\e[0;31m\] '

GUESTFISH_OUTPUT='\e[0m'
GUESTFISH_RESTORE="$GUESTFISH_OUTPUT"
GUESTFISH_INIT='\e[1;34m'

Blue welcome text, green prompt, red commands, black command output.

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
help -l|--list

Without any parameter, this provides general help.

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

With -l or --list, this list all commands.

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 "disk-create".

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

COMMANDS

acl-delete-def-file
acl-delete-def-file dir

This function deletes the default POSIX Access Control List (ACL) attached to directory "dir".

This command depends on the feature "acl". See also "feature-available".

acl-get-file
acl-get-file path acltype

This function returns the POSIX Access Control List (ACL) attached to "path". The ACL is returned in "long text form" (see acl(5)).

The "acltype" parameter may be:
"access"

Return the ordinary (access) ACL for any file, directory or other filesystem object.

"default"

Return the default ACL. Normally this only makes sense if "path" is a directory.

This command depends on the feature "acl". See also "feature-available".

acl-set-file
acl-set-file path acltype acl

This function sets the POSIX Access Control List (ACL) attached to "path".

The "acltype" parameter may be:
"access"

Set the ordinary (access) ACL for any file, directory or other filesystem object.

"default"

Set the default ACL. Normally this only makes sense if "path" is a directory.

The "acl" parameter is the new ACL in either "long text form" or "short text form" (see acl(5)). The new ACL completely replaces any previous ACL on the file. The ACL must contain the full Unix permissions (eg. "u::rwx,g::rx,o::rx").

If you are specifying individual users or groups, then the mask field is also required (eg. "m::rwx"), followed by the "u:ID:..." and/or "g:ID:..." field(s). A full ACL string might therefore look like this:

u::rwx,g::rwx,o::rwx,m::rwx,u:500:rwx,g:500:rwx
\ Unix permissions / \mask/ \ ACL /

You should use numeric UIDs and GIDs. To map usernames and groupnames to the correct numeric ID in the context of the guest, use the Augeas functions (see "aug-init").

This command depends on the feature "acl". See also "feature-available".

add-cdrom
add-cdrom filename

This function adds a virtual CD-ROM disk image to the guest.

The image is added as read-only drive, so this function is equivalent of "add-drive-ro".

This function is deprecated. In new code, use the "add-drive-ro" 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.

add-domain
domain
add-domain dom [libvirturi:..] [readonly:true|false] [iface:..] [live:true|false] [allowuuid:true|false] [readonlydisk:..] [cachemode:..] [discard:..] [copyonread:true|false]

This function adds the disk(s) attached to the named libvirt domain "dom". It works by connecting to libvirt, requesting the domain and domain XML from libvirt, parsing it for disks, and calling "add-drive-opts" on each one.

The number of disks added is returned. This operation is atomic: if an error is returned, then no disks are added.

This function does some minimal checks to make sure the libvirt domain is not running (unless "readonly" is true). In a future version we will try to acquire the libvirt lock on each disk.

Disks must be accessible locally. This often means that adding disks from a remote libvirt connection (see https://libvirt.org/remote.html) will fail unless those disks are accessible via the same device path locally too.

The optional "libvirturi" parameter sets the libvirt URI (see https://libvirt.org/uri.html). If this is not set then we connect to the default libvirt URI (or one set through an environment variable, see the libvirt documentation for full details).

The optional "live" flag is ignored in libguestfs ≥ 1.48.

If the "allowuuid" flag is true (default is false) then a UUID may be passed instead of the domain name. The "dom" string is treated as a UUID first and looked up, and if that lookup fails then we treat "dom" as a name as usual.

The optional "readonlydisk" parameter controls what we do for disks which are marked <readonly/> in the libvirt XML. Possible values are:
readonlydisk = "error"

If "readonly" is false:

The whole call is aborted with an error if any disk with the <readonly/> flag is found.

If "readonly" is true:

Disks with the <readonly/> flag are added read-only.

readonlydisk = "read"

If "readonly" is false:

Disks with the <readonly/> flag are added read-only. Other disks are added read/write.

If "readonly" is true:

Disks with the <readonly/> flag are added read-only.

readonlydisk = "write" (default)

If "readonly" is false:

Disks with the <readonly/> flag are added read/write.

If "readonly" is true:

Disks with the <readonly/> flag are added read-only.

readonlydisk = "ignore"

If "readonly" is true or false:

Disks with the <readonly/> flag are skipped.

If present, the value of "logical_block_size" attribute of <blockio/> tag in libvirt XML will be passed as "blocksize" parameter to "add-drive-opts".

The other optional parameters are passed directly through to "add-drive-opts".

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

add-drive
add
add-drive-opts
add-drive filename [readonly:true|false] [format:..] [iface:..] [name:..] [label:..] [protocol:..] [server:..] [username:..] [secret:..] [cachemode:..] [discard:..] [copyonread:true|false] [blocksize:N]

This function adds a disk image called filename to the handle. filename may be a regular host file or a host device.

When this function is called before "launch" (the usual case) then the first time you call this function, the disk appears in the API as /dev/sda, the second time as /dev/sdb, and so on.

You don’t necessarily need to be root when using libguestfs. However you obviously do need sufficient permissions to access the filename for whatever operations you want to perform (ie. read access if you just want to read the image or write access if you want to modify the image).

This call checks that filename exists.

filename may be the special string "/dev/null". See "NULL DISKS" in guestfs(3).

The optional arguments are:
"readonly"

If true then the image is treated as read-only. Writes are still allowed, but they are stored in a temporary snapshot overlay which is discarded at the end. The disk that you add is not modified.

"format"

This forces the image format. If you omit this (or use "add-drive" or "add-drive-ro") then the format is automatically detected. Possible formats include "raw" and "qcow2".

Automatic detection of the format opens you up to a potential security hole when dealing with untrusted raw-format images. See CVE-2010-3851 and RHBZ#642934. Specifying the format closes this security hole.

"iface"

This rarely-used option lets you emulate the behaviour of the deprecated "add-drive-with-if" call (q.v.)

"name"

This field used to be passed as a hint for guest inspection, but it is no longer used.

"label"

Give the disk a label. The label should be a unique, short string using only ASCII characters "[a-zA-Z]". As well as its usual name in the API (such as /dev/sda), the drive will also be named /dev/disk/guestfs/label.

See "DISK LABELS" in guestfs(3).

"protocol"

The optional protocol argument can be used to select an alternate source protocol.

See also: "REMOTE STORAGE" in guestfs(3).
"protocol = "file""

filename is interpreted as a local file or device. This is the default if the optional protocol parameter is omitted.

"protocol = "ftp"|"ftps"|"http"|"https"|"tftp""

Connect to a remote FTP, HTTP or TFTP server. The "server" parameter must also be supplied - see below.

See also: "FTP, HTTP AND TFTP" in guestfs(3)

"protocol = "gluster""

Connect to the GlusterFS server. The "server" parameter must also be supplied - see below.

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 mountable

This command tests whether "mountable" 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 https://wiki.osdev.org/ISO_9660#The_Primary_Volume_Descriptor

journal-close
journal-close

Close the journal handle.

This command depends on the feature "journal". See also "feature-available".

journal-get
journal-get

Read the current journal entry. This returns all the fields in the journal as a set of "(attrname, attrval)" pairs. The "attrname" is the field name (a string).

The "attrval" is the field value (a binary blob, often but not always a string). Please note that "attrval" is a byte array, not a \0-terminated C string.

The length of data may be truncated to the data threshold (see: "journal-set-data-threshold", "journal-get-data-threshold").

If you set the data threshold to unlimited (0) then this call can read a journal entry of any size, ie. it is not limited by the libguestfs protocol.

This command depends on the feature "journal". See also "feature-available".

journal-get-data-threshold
journal-get-data-threshold

Get the current data threshold for reading journal entries. This is a hint to the journal that it may truncate data fields to this size when reading them (note also that it may not truncate them). If this returns 0, then the threshold is unlimited.

EXIT STATUS

guestfish returns 0 if the commands completed without error, or 1 if there was an error.

ENVIRONMENT VARIABLES

EDITOR

The "edit" command uses $EDITOR as the editor. If not set, it uses "vi".

GUESTFISH_DISPLAY_IMAGE

The "display" command uses $GUESTFISH_DISPLAY_IMAGE to display images. If not set, it uses display(1).

GUESTFISH_INIT

Printed when guestfish starts. See "PROMPT".

GUESTFISH_OUTPUT

Printed before guestfish output. See "PROMPT".

GUESTFISH_PID

Used with the --remote option to specify the remote guestfish process to control. See section "REMOTE CONTROL GUESTFISH OVER A SOCKET".

GUESTFISH_PS1

Set the command prompt. See "PROMPT".

GUESTFISH_RESTORE

Printed before guestfish exits. See "PROMPT".

HEXEDITOR

The "hexedit" command uses $HEXEDITOR as the external hex editor. If not specified, the external hexedit(1) program is used.

HOME

If compiled with GNU readline support, various files in the home directory can be used. See "FILES".

LIBGUESTFS_APPEND

Pass additional options to the guest kernel.

LIBGUESTFS_ATTACH_METHOD

This is the old way to set "LIBGUESTFS_BACKEND".

LIBGUESTFS_BACKEND

Choose the default way to create the appliance. See "guestfs_set_backend" in guestfs(3).

LIBGUESTFS_BACKEND_SETTINGS

A colon-separated list of backend-specific settings. See "BACKEND" in guestfs(3), "BACKEND SETTINGS" in guestfs(3).

LIBGUESTFS_CACHEDIR

The location where libguestfs will cache its appliance, when using a supermin appliance. The appliance is cached and shared between all handles which have the same effective user ID.

If "LIBGUESTFS_CACHEDIR" is not set, then "TMPDIR" is used. If "TMPDIR" is not set, then /var/tmp is used.

See also "LIBGUESTFS_TMPDIR", "set-cachedir".

LIBGUESTFS_DEBUG

Set "LIBGUESTFS_DEBUG=1" to enable verbose messages. This has the same effect as using the -v option.

LIBGUESTFS_HV

Set the default hypervisor (usually qemu) binary that libguestfs uses. If not set, then the qemu which was found at compile time by the configure script is used.

LIBGUESTFS_MEMSIZE

Set the memory allocated to the qemu process, in megabytes. For example:

LIBGUESTFS_MEMSIZE=700

LIBGUESTFS_PATH

Set the path that guestfish uses to search for kernel and initrd.img. See the discussion of paths in guestfs(3).

LIBGUESTFS_QEMU

This is the old way to set "LIBGUESTFS_HV".

LIBGUESTFS_TMPDIR

The location where libguestfs will store temporary files used by each handle.

If "LIBGUESTFS_TMPDIR" is not set, then "TMPDIR" is used. If "TMPDIR" is not set, then /tmp is used.

See also "LIBGUESTFS_CACHEDIR", "set-tmpdir".

LIBGUESTFS_TRACE

Set "LIBGUESTFS_TRACE=1" to enable command traces.

PAGER

The "more" command uses $PAGER as the pager. If not set, it uses "more".

PATH

Libguestfs and guestfish may run some external programs, and rely on $PATH being set to a reasonable value. If using the libvirt backend, libvirt will not work at all unless $PATH contains the path of qemu/KVM.

SUPERMIN_KERNEL
SUPERMIN_KERNEL_VERSION
SUPERMIN_MODULES

These three environment variables allow the kernel that libguestfs uses in the appliance to be selected. If $SUPERMIN_KERNEL is not set, then the most recent host kernel is chosen. For more information about kernel selection, see supermin(1).

TMPDIR

See "LIBGUESTFS_CACHEDIR", "LIBGUESTFS_TMPDIR".

XDG_RUNTIME_DIR

This directory represents a user-specific directory for storing non-essential runtime files.

If it is set, then is used to store temporary sockets and PID files. Otherwise, /tmp is used.

FILES

$XDG_CONFIG_HOME/libguestfs/libguestfs-tools.conf
$HOME/.libguestfs-tools.rc
$XDG_CONFIG_DIRS/libguestfs/libguestfs-tools.conf
/etc/libguestfs-tools.conf

This configuration file controls the default read-only or read-write mode (--ro or --rw).

See libguestfs-tools.conf(5).

$HOME/.guestfish

If compiled with GNU readline support, then the command history is saved in this file.

$HOME/.inputrc
/etc/inputrc

If compiled with GNU readline support, then these files can be used to configure readline. For further information, please see "INITIALIZATION FILE" in readline(3).

To write rules which only apply to guestfish, use:

$if guestfish
...
$endif

Variables that you can set in inputrc that change the behaviour of guestfish in useful ways include:
completion-ignore-case (default: on)

By default, guestfish will ignore case when tab-completing paths on the disk. Use:

set completion-ignore-case off

to make guestfish case sensitive.

test1.img
test2.img (etc)

When using the -N or --new option, the prepared disk or filesystem will be created in the file test1.img in the current directory. The second use of -N will use test2.img and so on. Any existing file with the same name will be overwritten. You can use a different filename by using the "filename=" prefix.

AUTHORS

Richard W.M. Jones ("rjones at redhat dot com")

COPYRIGHT

LICENSE

This program is free software; you can redistribute it and/or modify it under the terms of the GNU General Public License as published by the Free Software Foundation; either version 2 of the License, or (at your option) any later version.

This program is distributed in the hope that it will be useful, but WITHOUT ANY WARRANTY; without even the implied warranty of MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the GNU General Public License for more details.

You should have received a copy of the GNU General Public License along with this program; if not, write to the Free Software Foundation, Inc., 51 Franklin Street, Fifth Floor, Boston, MA 02110-1301 USA.

BUGS

To get a list of bugs against libguestfs, use this link: https://bugzilla.redhat.com/buglist.cgi?component=libguestfs&product=Virtualization+Tools

To report a new bug against libguestfs, use this link: https://bugzilla.redhat.com/enter_bug.cgi?component=libguestfs&product=Virtualization+Tools

When reporting a bug, please supply:

	•		The version of libguestfs.
	•		Where you got libguestfs (eg. which Linux distro, compiled from source, etc)
	•		Describe the bug accurately and give a way to reproduce it.
	•		Run libguestfs-test-tool(1) and paste the complete, unedited output into the bug report.

	’b’		Block special
	’c’		Char special
	’d’		Directory
	’f’		FIFO (named pipe)
	’l’		Symbolic link
	’r’		Regular file
	’s’		Socket
	’h’		Shadow inode (Solaris)
	’w’		Whiteout inode (BSD)
	’u’		Unknown file type

	’A’		When the file is accessed, its atime is not modified.
	’a’		The file is append-only.
	’c’		The file is compressed on-disk.
	’D’		(Directories only.) Changes to this directory are written synchronously to disk.
	’d’		The file is not a candidate for backup (see dump(8)).
	’E’		The file has compression errors.
	’e’		The file is using extents.
	’h’		The file is storing its blocks in units of the filesystem blocksize instead of sectors.
	’I’		(Directories only.) The directory is using hashed trees.
	’i’		The file is immutable. It cannot be modified, deleted or renamed. No link can be created to this file.
	’j’		The file is data-journaled.
	’s’		When the file is deleted, all its blocks will be zeroed.
	’S’		Changes to this file are written synchronously to disk.
	’T’		(Directories only.) This is a hint to the block allocator that subdirectories contained in this directory should be spread across blocks. If not present, the block allocator will try to group subdirectories together.
	’t’		For a file, this disables tail-merging. (Not used by upstream implementations of ext2.)
	’u’		When the file is deleted, its blocks will be saved, allowing the file to be undeleted.
	’X’		The raw contents of the compressed file may be accessed.
	’Z’		The compressed file is dirty.

	"W"		write-mostly
	"F"		device is faulty
	"S"		device is a RAID spare
	"R"		replacement

	’b’		Block special
	’c’		Char special
	’d’		Directory
	’f’		FIFO (named pipe)
	’l’		Symbolic link
	’r’		Regular file
	’s’		Socket
	’u’		Unknown file type
	’?’		The readdir(3) call returned a "d_type" field with an unexpected value

Manpages

NAME

SYNOPSIS

WARNING

DESCRIPTION

EXAMPLES

OPTIONS

COMMANDS ON COMMAND LINE

USING launch (OR run)

OPENING DISKS FOR READ AND WRITE

QUOTING

OPTIONAL ARGUMENTS

NUMBERS

WILDCARDS AND GLOBBING

COMMENTS

RUNNING COMMANDS LOCALLY

PIPES

HOME DIRECTORIES

ENCRYPTED DISKS

WINDOWS PATHS

UPLOADING AND DOWNLOADING FILES

EXIT ON ERROR BEHAVIOUR

REMOTE CONTROL GUESTFISH OVER A SOCKET

PREPARED DISK IMAGES

ADDING REMOTE STORAGE

PROGRESS BARS

PROMPT

WINDOWS 8

GUESTFISH COMMANDS

COMMANDS

EXIT STATUS

ENVIRONMENT VARIABLES

FILES

SEE ALSO

AUTHORS

COPYRIGHT

LICENSE

BUGS

	•		"feature-available" is the same as this call, but with a slightly simpler to use API: that call returns a boolean true/false instead of throwing an error.
	•		You must call "launch" before calling this function.

	•		If a group of functions is available, this does not necessarily mean that they will work. You still have to check for errors when calling individual API functions even if they are available.
	•		It is usually the job of distro packagers to build complete functionality into the libguestfs appliance. Upstream libguestfs, if built from source with all requirements satisfied, will support everything.
	•		This call was added in version 1.0.80. In previous versions of libguestfs all you could do would be to speculatively execute a command to find out if the daemon implemented it. See also "version".

	•		many types of Un*x and Linux binary
	•		many types of Un*x and Linux shared library
	•		Windows Win32 and Win64 binaries
	•		Windows Win32 and Win64 DLLs

	•		Linux kernel modules
	•		Linux new-style initrd images
	•		some non-x86 Linux vmlinuz kernels

	•		static libraries (libfoo.a)
	•		Linux old-style initrd as compressed ext2 filesystem (RHEL 3)
	•		x86 Linux vmlinuz kernels

	•		The resulting list is written to an external file.
	•		Items (filenames) in the result are separated by "\0" characters. See find(1) option -print0.
	•		The result list is not sorted.

	•		Multiple status codes can be summed together.
	•		A non-zero return code can mean "success", for example if errors have been corrected on the filesystem.
	•		Checking or repairing NTFS volumes is not supported (by linux-ntfs).

	•		There is currently no way in the API to install grub2, which is used by most modern Linux guests. It is possible to run the grub2 command from the guest, although see the caveats in "RUNNING COMMANDS" in guestfs(3).
	•		This uses grub-install(8) from the host. Unfortunately grub is not always compatible with itself, so this only works in rather narrow circumstances. Careful testing with each guest version is advisable.
	•		If grub-install reports the error "No suitable drive was found in the generated device map." it may be that you need to create a /boot/grub/device.map file first that contains the mapping between grub device names and Linux device names. It is usually sufficient to create a file containing:

	•		Unlike most other inspection API calls, the guest’s disks must be mounted up before you call this, since it needs to read information from the guest filesystem during the call.
	•		Security: The icon data comes from the untrusted guest, and should be treated with caution. PNG files have been known to contain exploits. Ensure that libpng (or other relevant libraries) are fully up to date before trying to process or display the icon.
	•		The PNG image returned can be any size. It might not be square. Libguestfs tries to return the largest, highest quality icon available. The application must scale the icon to the required size.
	•		Extracting icons from Windows guests requires the external wrestool(1) program from the "icoutils" package, and several programs (bmptopnm(1), pnmtopng(1), pamcut(1)) from the "netpbm" package. These must be installed separately.
	•		Operating system icons are usually trademarks. Seek legal advice before using trademarks in applications.

	s		Single 7-bit-byte characters like ASCII and the ASCII-compatible parts of ISO-8859-X (this is what "strings" uses).
	S		Single 8-bit-byte characters.
	b		16-bit big endian strings such as those encoded in UTF-16BE or UCS-2BE.

	B		32-bit big endian such as UCS-4BE.
	L		32-bit little endian such as UCS-4LE.