guestfish - the guest filesystem shell
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 -iUsing 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.
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.
$ 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> exitCreate 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_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.confUse 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/groupAnother way to edit /boot/grub/grub.conf interactively is:
guestfish --rw -a disk.img -i edit /boot/grub/grub.confCreate 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/sda1An alternate way to create a 100MB disk called test1.img containing a single ext2-formatted partition:
guestfish -N fsTo list what is available do:
guestfish -N help | lessAccess a remote disk using ssh:
guestfish -a ssh://example.com/path/to/disk.img eval "`guestfish --listen`"
guestfish --remote add-ro disk.img
guestfish --remote run
guestfish --remote lvsDisplays general help on options.
Lists all available guestfish commands.
Displays detailed help on a single command cmd.
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.
Add a remote disk. See "ADDING REMOTE STORAGE".
When used in conjunction with the -d option, this specifies the libvirt URI to use. The default is to use the default libvirt connection.
If using the --listen option and a csh-like shell, use this option. See section "REMOTE CONTROL AND CSH" below.
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.
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.
Read commands from file. To write pure guestfish scripts, use:
#!/usr/bin/guestfish -fThe 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.imgforces raw format (no auto-detection) for disk.img.
guestfish --format=raw -a disk.img --format -a another.imgforces 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".
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 -iNote 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-domainUsing this flag is mostly equivalent to using the inspect-os command and then using other commands to mount the filesystems that were found.
Read key or passphrase parameters from stdin. The default is to try to read passphrases from the user by opening /dev/tty.
Fork into the background and listen for remote commands. See section "REMOTE CONTROL GUESTFISH OVER A SOCKET" below.
Connect to a live virtual machine. (Experimental, see "ATTACHING TO RUNNING DAEMONS" in guestfs(3)).
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_xattrUsing 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.
Enable QEMU user networking in the guest.
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.
Disable autosync. This is enabled by default. See the discussion of autosync in the guestfs(3) manpage.
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.
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 | headdoesn't give an error.
Enable progress bars, even when guestfish is used non-interactively.
Progress bars are enabled by default when guestfish is used as an interactive shell.
Disable progress bars.
Send remote commands to $GUESTFISH_PID or pid. See section "REMOTE CONTROL GUESTFISH OVER A SOCKET" below.
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.
Enable SELinux support for the guest. See "SELINUX" in guestfs(3).
Enable very verbose messages. This is particularly useful if you find a bug.
Display the guestfish / libguestfs version number and exit.
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.
Echo each command before executing it.
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.
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.
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.
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\'"In double-quoted arguments (only) use backslash to insert special characters:
\aAlert (bell) character.
\bBackspace character.
\fForm feed character.
\nNewline character.
\rCarriage return character.
\tHorizontal tab character.
\vVertical tab character.
\"A literal double quote character.
\oooA character with octal value ooo. There must be precisely 3 octal digits (unlike C).
\xhhA 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.
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:falseEach optional argument can appear at most once. All optional arguments must appear after the required ones.
This section applies to all commands which can take integers as parameters.
When the command takes a parameter measured in bytes, you can use one of the following suffixes to specify kilobytes, megabytes and larger sizes:
The size in kilobytes (multiplied by 1024).
The size in SI 1000 byte units.
The size in megabytes (multiplied by 1048576).
The size in SI 1000000 byte units.
The size in gigabytes (multiplied by 2**30).
The size in SI 10**9 byte units.
The size in terabytes (multiplied by 2**40).
The size in SI 10**12 byte units.
The size in petabytes (multiplied by 2**50).
The size in SI 10**15 byte units.
The size in exabytes (multiplied by 2**60).
The size in SI 10**18 byte units.
The size in zettabytes (multiplied by 2**70).
The size in SI 10**21 byte units.
The size in yottabytes (multiplied by 2**80).
The size in SI 10**24 byte units.
For example:
truncate-size /file 1Gwould 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.
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 1234When 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).
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/maryglob 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.
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 commentBlank lines are also ignored.
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.gzwill 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.
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; doneor with names like /foo.001:
<! for n in `seq 1 100`; do printf "write /foo.%03d %d\n" $n $n; doneWhen 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.
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 "|"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 "~"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_LUKSThen 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 trueThe 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/luksdevIf 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.LOGThe 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/barThis only works in argument positions that expect a path.
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 - /fooreads 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
ENDAny 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).
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.
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 exitNote 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".
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=$pid2 cmdWhen using csh-like shells (csh, tcsh etc) you have to add the --csh option:
eval "`guestfish --listen --csh`"Remote control happens over a Unix domain socket called /tmp/.guestfish-$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.
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
# ...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.imgDon't use these options. Use the equivalent commands instead, eg:
guestfish --remote add-drive disk.imgor:
guestfish --remote
><fs> add disk.imgUsing 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.
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 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.
Create a 100MB disk with an ext4-formatted partition, called test1.img in the current directory:
guestfish -N fs:ext4Create a 32MB disk with a VFAT-formatted partition, and mount it:
guestfish -N fs:vfat:32M -m /dev/sda1Create a blank 200MB disk:
guestfish -N disk:200MCreate a blank 200MB disk called blankdisk.img (instead of test1.img):
guestfish -N blankdisk.img=disk:200Mguestfish -N [filename=]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 imageguestfish -N [filename=]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 typeguestfish -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 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 typeguestfish -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 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 typeguestfish -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 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 typeguestfish -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 100M the size of the disk image
bootsize 32M the size of the boot filesystem
partition mbr partition table typeguestfish -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 100M the size of the disk image
bootsize 32M the size of the boot filesystem
partition mbr partition table typeFor 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.imgURIs 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:rootThe possible -a URI formats are described below.
Add the local disk image (or device) called 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.comAdd 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 /disk protocol:gluster server:tcp:example.comAdd a disk located on an iSCSI server.
The equivalent API command would be:
><fs> add /target-iqn-name/lun protocol:iscsi server:tcp:example.comAdd 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]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 /disk protocol:rbd server:tcp:example.comAdd 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 /disk protocol:sheepdog [server:tcp:example.com]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]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% [#####-----------------------------------------] 00:30The 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.
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 bell character.
An ASCII ESC (escape) character.
A newline.
A carriage return.
The ASCII character whose code is the octal value NNN.
The ASCII character whose code is the hex value NN.
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 "fast startup" can prevent guestfish from mounting NTFS partitions. See "WINDOWS HIBERNATION AND WINDOWS 8 FAST STARTUP" in guestfs(3).
The commands in this section are guestfish convenience commands, in other words, they are not part of the guestfs(3) API.
help
help cmdWithout any parameter, this provides general help.
With a cmd parameter, this displays detailed help for that command.
This exits guestfish. You can also use ^D key.
alloc filename sizeThis 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 local [local ...] /remotedircopy-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 remote [remote ...] localdircopy-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 nameDelete 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 filenameUse 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=eogSee also display(1).
echo [params ...]This echos the parameters to the terminal.
edit filenameThis is used to edit a file. It downloads the file, edits it locally using your editor, then uploads the result.
The editor is $EDITOR. However if you use the alternate commands vi or emacs you will get those corresponding editors.
event name eventset "shell script ..."Register a shell script fragment which is executed when an event is raised. See "guestfs_set_event_callback" in guestfs(3) for a discussion of the event API in libguestfs.
The name parameter is a name that you give to this event handler. It can be any string (even the empty string) and is simply there so you can delete the handler using the guestfish delete-event command.
The eventset parameter is a comma-separated list of one or more events, for example close or close,trace. The special value * means all events.
The third and final parameter is the shell script fragment (or any external command) that is executed when any of the events in the eventset occurs. It is executed using $SHELL -c, or if $SHELL is not set then /bin/sh -c.
The shell script fragment receives callback parameters as arguments $1, $2 etc. The actual event that was called is available in the environment variable $EVENT.
event "" close "echo closed"
event messages appliance,library,trace "echo $@"
event "" progress "echo progress: $3/$4"
event "" * "echo $EVENT $@"See also the guestfish commands delete-event and list-events.
glob command args...Expand wildcards in any paths in the args list, and run command repeatedly on each matching path.
hexedit <filename|device>
hexedit <filename|device> <max>
hexedit <filename|device> <start> <max>Use hexedit (a hex editor) to edit all or part of a binary file or block device.
This command works by downloading potentially the whole file or device, editing it locally, then uploading it. If the file or device is large, you have to specify which part you wish to edit by using max and/or start max parameters. start and max are specified in bytes, with the usual modifiers allowed such as 1M (1 megabyte).
For example to edit the first few sectors of a disk you might do:
hexedit /dev/sda 1Mwhich would allow you to edit anywhere within the first megabyte of the disk.
To edit the superblock of an ext2 filesystem on /dev/sda1, do:
hexedit /dev/sda1 0x400 0x400(assuming the superblock is in the standard location).
This command requires the external hexedit(1) program. You can specify another program to use by setting the HEXEDITOR environment variable.
See also "hexdump".
lcd directoryChange the local directory, ie. the current directory of guestfish itself.
Note that !cd won't do what you might expect.
list-eventsList the event handlers registered using the guestfish event command.
manOpens the manual page for guestfish.
more filename
less filenameThis is used to view a file.
The default viewer is $PAGER. However if you use the alternate command less you will get the less command specifically.
reopenClose and reopen the libguestfs handle. It is not necessary to use this normally, because the handle is closed properly when guestfish exits. However this is occasionally useful for testing.
setenv VAR valueSet the environment variable VAR to the string value.
To print the value of an environment variable use a shell command such as:
!echo $VAR sparse filename sizeThis creates an empty sparse file of the given size, and then adds so it can be further examined.
In all respects it works the same as the "alloc" command, except that the image file is allocated sparsely, which means that disk blocks are not assigned to the file until they are needed. Sparse disk files only use space when written to, but they are slower and there is a danger you could run out of real disk space during a write operation.
For more advanced image creation, see qemu-img(1) utility.
Size can be specified using standard suffixes, eg. 1M.
See also the guestfish "scratch" command.
supportedThis command returns a list of the optional groups known to the daemon, and indicates which ones are supported by this build of the libguestfs appliance.
See also "AVAILABILITY" in guestfs(3).
time command args...Run the command as usual, but print the elapsed time afterwards. This can be useful for benchmarking operations.
unsetenv VARRemove VAR from the environment.
acl-delete-def-file dirThis function deletes the default POSIX Access Control List (ACL) attached to directory dir.
acl-get-file path acltypeThis 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:
accessReturn the ordinary (access) ACL for any file, directory or other filesystem object.
defaultReturn the default ACL. Normally this only makes sense if path is a directory.
acl-set-file path acltype aclThis function sets the POSIX Access Control List (ACL) attached to path.
The acltype parameter may be:
accessSet the ordinary (access) ACL for any file, directory or other filesystem object.
defaultSet 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").
add-cdrom filenameThis function adds a virtual CD-ROM disk image to the guest.
Do not use this function! ISO files are just ordinary read-only disk images. Use "add-drive-ro" instead.
This function is deprecated. In new code, use the "add-drive" 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 dom [libvirturi:..] [readonly:true|false] [iface:..] [live:true|false] [allowuuid:true|false] [readonlydisk:..]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 http://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 http://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 controls whether this call will try to connect to a running virtual machine guestfsd process if it sees a suitable <channel> element in the libvirt XML definition. The default (if the flag is omitted) is never to try. See "ATTACHING TO RUNNING DAEMONS" in guestfs(3) for more information.
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:
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.
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.
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.
If readonly is true or false:
Disks with the <readonly/> flag are skipped.
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 filename [readonly:true|false] [format:..] [iface:..] [name:..] [label:..] [protocol:..] [server:..] [username:..] [secret:..] [cachemode:..]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.
In libguestfs ≥ 1.20 you can also call this function after launch (with some restrictions). This is called "hotplugging". When hotplugging, you must specify a label so that the new disk gets a predictable name. For more information see "HOTPLUGGING" in guestfs(3).
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:
readonlyIf 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.
formatThis 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.
ifaceThis rarely-used option lets you emulate the behaviour of the deprecated "add-drive-with-if" call (q.v.)
nameThe name the drive had in the original guest, e.g. /dev/sdb. This is used as a hint to the guest inspection process if it is available.
labelGive 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.
protocolThe 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: "GLUSTER" in guestfs(3)
protocol = "iscsi"Connect to the iSCSI server. The server parameter must also be supplied - see below.
See also: "ISCSI" in guestfs(3).
protocol = "nbd"Connect to the Network Block Device server. The server parameter must also be supplied - see below.
See also: "NETWORK BLOCK DEVICE" in guestfs(3).
protocol = "rbd"Connect to the Ceph (librbd/RBD) server. The server parameter must also be supplied - see below. The username parameter may be supplied. See below. The secret parameter may be supplied. See below.
See also: "CEPH" in guestfs(3).
protocol = "sheepdog"Connect to the Sheepdog server. The server parameter may also be supplied - see below.
See also: "SHEEPDOG" in guestfs(3).
protocol = "ssh"Connect to the Secure Shell (ssh) server.
The server parameter must be supplied. The username parameter may be supplied. See below.
See also: "SSH" in guestfs(3).
serverFor protocols which require access to a remote server, this is a list of server(s).
Protocol Number of servers required
-------- --------------------------
file List must be empty or param not used at all
ftp|ftps|http|https|tftp Exactly one
gluster Exactly one
iscsi Exactly one
nbd Exactly one
rbd One or more
sheepdog Zero or more
ssh Exactly oneEach list element is a string specifying a server. The string must be in one of the following formats:
hostname
hostname:port
tcp:hostname
tcp:hostname:port
unix:/path/to/socketIf the port number is omitted, then the standard port number for the protocol is used (see /etc/services).
usernameFor the ftp, ftps, http, https, iscsi, rbd, ssh and tftp protocols, this specifies the remote username.
If not given, then the local username is used for ssh, and no authentication is attempted for ceph. But note this sometimes may give unexpected results, for example if using the libvirt backend and if the libvirt backend is configured to start the qemu appliance as a special user such as qemu.qemu. If in doubt, specify the remote username you want.
secretFor the rbd protocol only, this specifies the 'secret' to use when connecting to the remote device.
If not given, then a secret matching the given username will be looked up in the default keychain locations, or if no username is given, then no authentication will be used.
cachemodeChoose whether or not libguestfs will obey sync operations (safe but slow) or not (unsafe but fast). The possible values for this string are:
cachemode = "writeback"This is the default.
Write operations in the API do not return until a write(2) call has completed in the host [but note this does not imply that anything gets written to disk].
Sync operations in the API, including implicit syncs caused by filesystem journalling, will not return until an fdatasync(2) call has completed in the host, indicating that data has been committed to disk.
cachemode = "unsafe"In this mode, there are no guarantees. Libguestfs may cache anything and ignore sync requests. This is suitable only for scratch or temporary disks.
This command has one or more optional arguments. See "OPTIONAL ARGUMENTS".
add-drive-ro filenameThis function is the equivalent of calling "add-drive-opts" with the optional parameter GUESTFS_ADD_DRIVE_OPTS_READONLY set to 1, so the disk is added read-only, with the format being detected automatically.
add-drive-ro-with-if filename ifaceThis is the same as "add-drive-ro" but it allows you to specify the QEMU interface emulation to use at run time.
This function is deprecated. In new code, use the "add-drive" 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-drive-scratch size [name:..] [label:..]This command adds a temporary scratch drive to the handle. The size parameter is the virtual size (in bytes). The scratch drive is blank initially (all reads return zeroes until you start writing to it). The drive is deleted when the handle is closed.
The optional arguments name and label are passed through to "add-drive".
This command has one or more optional arguments. See "OPTIONAL ARGUMENTS".
add-drive-with-if filename ifaceThis is the same as "add-drive" but it allows you to specify the QEMU interface emulation to use at run time.
This function is deprecated. In new code, use the "add-drive" 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.
aug-clear augpathSet the value associated with path to NULL. This is the same as the augtool(1) clear command.
aug-closeClose the current Augeas handle and free up any resources used by it. After calling this, you have to call "aug-init" again before you can use any other Augeas functions.
aug-defnode name expr valDefines a variable name whose value is the result of evaluating expr.
If expr evaluates to an empty nodeset, a node is created, equivalent to calling "aug-set" expr, value. name will be the nodeset containing that single node.
On success this returns a pair containing the number of nodes in the nodeset, and a boolean flag if a node was created.
aug-defvar name exprDefines an Augeas variable name whose value is the result of evaluating expr. If expr is NULL, then name is undefined.
On success this returns the number of nodes in expr, or 0 if expr evaluates to something which is not a nodeset.
aug-get augpathLook up the value associated with path. If path matches exactly one node, the value is returned.
aug-init root flagsCreate a new Augeas handle for editing configuration files. If there was any previous Augeas handle associated with this guestfs session, then it is closed.
You must call this before using any other "aug-*" commands.
root is the filesystem root. root must not be NULL, use / instead.
The flags are the same as the flags defined in <augeas.h>, the logical or of the following integers:
AUG_SAVE_BACKUP = 1Keep the original file with a .augsave extension.
AUG_SAVE_NEWFILE = 2Save changes into a file with extension .augnew, and do not overwrite original. Overrides AUG_SAVE_BACKUP.
AUG_TYPE_CHECK = 4Typecheck lenses.
This option is only useful when debugging Augeas lenses. Use of this option may require additional memory for the libguestfs appliance. You may need to set the LIBGUESTFS_MEMSIZE environment variable or call "set-memsize".
AUG_NO_STDINC = 8Do not use standard load path for modules.
AUG_SAVE_NOOP = 16Make save a no-op, just record what would have been changed.
AUG_NO_LOAD = 32Do not load the tree in "aug-init".
To close the handle, you can call "aug-close".
To find out more about Augeas, see http://augeas.net/.
aug-insert augpath label true|falseCreate a new sibling label for path, inserting it into the tree before or after path (depending on the boolean flag before).
path must match exactly one existing node in the tree, and label must be a label, ie. not contain /, * or end with a bracketed index [N].
aug-label augpathThe label (name of the last element) of the Augeas path expression augpath is returned. augpath must match exactly one node, else this function returns an error.
aug-loadLoad files into the tree.
See aug_load in the Augeas documentation for the full gory details.
aug-ls augpathThis is just a shortcut for listing "aug-match" path/* and sorting the resulting nodes into alphabetical order.
aug-match augpathReturns a list of paths which match the path expression path. The returned paths are sufficiently qualified so that they match exactly one node in the current tree.
aug-mv src destMove the node src to dest. src must match exactly one node. dest is overwritten if it exists.
aug-rm augpathRemove path and all of its children.
On success this returns the number of entries which were removed.
aug-saveThis writes all pending changes to disk.
The flags which were passed to "aug-init" affect exactly how files are saved.
aug-set augpath valSet the value associated with path to val.
In the Augeas API, it is possible to clear a node by setting the value to NULL. Due to an oversight in the libguestfs API you cannot do that with this call. Instead you must use the "aug-clear" call.
aug-setm base sub valChange multiple Augeas nodes in a single operation. base is an expression matching multiple nodes. sub is a path expression relative to base. All nodes matching base are found, and then for each node, sub is changed to val. sub may also be NULL in which case the base nodes are modified.
This returns the number of nodes modified.
available 'groups ...'This command is used to check the availability of some groups of functionality in the appliance, which not all builds of the libguestfs appliance will be able to provide.
The libguestfs groups, and the functions that those groups correspond to, are listed in "AVAILABILITY" in guestfs(3). You can also fetch this list at runtime by calling "available-all-groups".
The argument groups is a list of group names, eg: ["inotify", "augeas"] would check for the availability of the Linux inotify functions and Augeas (configuration file editing) functions.
The command returns no error if all requested groups are available.
It fails with an error if one or more of the requested groups is unavailable in the appliance.
If an unknown group name is included in the list of groups then an error is always returned.
Notes:
"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.
The reason is because we don't know what groups are supported by the appliance/daemon until it is running and can be queried.
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".
See also "filesystem-available".
available-all-groupsThis command returns a list of all optional groups that this daemon knows about. Note this returns both supported and unsupported groups. To find out which ones the daemon can actually support you have to call "available" / "feature-available" on each member of the returned list.
See also "available", "feature-available" and "AVAILABILITY" in guestfs(3).
base64-in (base64file|-) filenameThis command uploads base64-encoded data from base64file to filename.
Use - instead of a filename to read/write from stdin/stdout.
base64-out filename (base64file|-)This command downloads the contents of filename, writing it out to local file base64file encoded as base64.
Use - instead of a filename to read/write from stdin/stdout.
blkid deviceThis command returns block device attributes for device. The following fields are usually present in the returned hash. Other fields may also be present.
UUIDThe uuid of this device.
LABELThe label of this device.
VERSIONThe version of blkid command.
TYPEThe filesystem type or RAID of this device.
USAGEThe usage of this device, for example filesystem or raid.
blockdev-flushbufs deviceThis tells the kernel to flush internal buffers associated with device.
This uses the blockdev(8) command.
blockdev-getbsz deviceThis returns the block size of a device.
Note: this is different from both size in blocks and filesystem block size. Also this setting is not really used by anything. You should probably not use it for anything. Filesystems have their own idea about what block size to choose.
This uses the blockdev(8) command.
blockdev-getro deviceReturns a boolean indicating if the block device is read-only (true if read-only, false if not).
This uses the blockdev(8) command.
blockdev-getsize64 deviceThis returns the size of the device in bytes.
See also "blockdev-getsz".
This uses the blockdev(8) command.
blockdev-getss deviceThis returns the size of sectors on a block device. Usually 512, but can be larger for modern devices.
(Note, this is not the size in sectors, use "blockdev-getsz" for that).
This uses the blockdev(8) command.
blockdev-getsz deviceThis returns the size of the device in units of 512-byte sectors (even if the sectorsize isn't 512 bytes ... weird).
See also "blockdev-getss" for the real sector size of the device, and "blockdev-getsize64" for the more useful size in bytes.
This uses the blockdev(8) command.
blockdev-rereadpt deviceReread the partition table on device.
This uses the blockdev(8) command.
blockdev-setbsz device blocksizeThis call does nothing and has never done anything because of a bug in blockdev. Do not use it.
If you need to set the filesystem block size, use the blocksize option of "mkfs".
This function is deprecated. In new code, use the "mkfs" 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.
blockdev-setro deviceSets the block device named device to read-only.
This uses the blockdev(8) command.
blockdev-setrw deviceSets the block device named device to read-write.
This uses the blockdev(8) command.
btrfs-device-add 'devices ...' fsAdd the list of device(s) in devices to the btrfs filesystem mounted at fs. If devices is an empty list, this does nothing.
btrfs-device-delete 'devices ...' fsRemove the devices from the btrfs filesystem mounted at fs. If devices is an empty list, this does nothing.
btrfs-filesystem-balance fsBalance the chunks in the btrfs filesystem mounted at fs across the underlying devices.
btrfs-filesystem-resize mountpoint [size:N]This command resizes a btrfs filesystem.
Note that unlike other resize calls, the filesystem has to be mounted and the parameter is the mountpoint not the device (this is a requirement of btrfs itself).
The optional parameters are:
sizeThe new size (in bytes) of the filesystem. If omitted, the filesystem is resized to the maximum size.
See also btrfs(8).
This command has one or more optional arguments. See "OPTIONAL ARGUMENTS".
btrfs-filesystem-sync fsForce sync on the btrfs filesystem mounted at fs.
btrfs-fsck device [superblock:N] [repair:true|false]Used to check a btrfs filesystem, device is the device file where the filesystem is stored.
This command has one or more optional arguments. See "OPTIONAL ARGUMENTS".
btrfs-set-seeding device true|falseEnable or disable the seeding feature of a device that contains a btrfs filesystem.
btrfs-subvolume-create destCreate a btrfs subvolume. The dest argument is the destination directory and the name of the snapshot, in the form /path/to/dest/name.
btrfs-subvolume-delete subvolumeDelete the named btrfs subvolume.
btrfs-subvolume-list fsList the btrfs snapshots and subvolumes of the btrfs filesystem which is mounted at fs.
btrfs-subvolume-set-default id fsSet the subvolume of the btrfs filesystem fs which will be mounted by default. See "btrfs-subvolume-list" to get a list of subvolumes.
btrfs-subvolume-snapshot source destCreate a writable snapshot of the btrfs subvolume source. The dest argument is the destination directory and the name of the snapshot, in the form /path/to/dest/name.
canonical-device-name deviceThis utility function is useful when displaying device names to the user. It takes a number of irregular device names and returns them in a consistent format:
/dev/hdX/dev/vdXThese are returned as /dev/sdX. Note this works for device names and partition names. This is approximately the reverse of the algorithm described in "BLOCK DEVICE NAMING" in guestfs(3).
/dev/mapper/VG-LV/dev/dm-NConverted to /dev/VG/LV form using "lvm-canonical-lvm-name".
Other strings are returned unmodified.
cap-get-file pathThis function returns the Linux capabilities attached to path. The capabilities set is returned in text form (see cap_to_text(3)).
If no capabilities are attached to a file, an empty string is returned.
cap-set-file path capThis function sets the Linux capabilities attached to path. The capabilities set cap should be passed in text form (see cap_from_text(3)).
case-sensitive-path pathThis can be used to resolve case insensitive paths on a filesystem which is case sensitive. The use case is to resolve paths which you have read from Windows configuration files or the Windows Registry, to the true path.
The command handles a peculiarity of the Linux ntfs-3g filesystem driver (and probably others), which is that although the underlying filesystem is case-insensitive, the driver exports the filesystem to Linux as case-sensitive.
One consequence of this is that special directories such as c:\windows may appear as /WINDOWS or /windows (or other things) depending on the precise details of how they were created. In Windows itself this would not be a problem.
Bug or feature? You decide: http://www.tuxera.com/community/ntfs-3g-faq/#posixfilenames1
"case-sensitive-path" attempts to resolve the true case of each element in the path. It will return a resolved path if either the full path or its parent directory exists. If the parent directory exists but the full path does not, the case of the parent directory will be correctly resolved, and the remainder appended unmodified. For example, if the file "/Windows/System32/netkvm.sys" exists:
"Windows/System32/netkvm.sys"
"Windows/System32/NoSuchFile"
ERROR
Note: Because of the above behaviour, "case-sensitive-path" cannot be used to check for the existence of a file.
Note: This function does not handle drive names, backslashes etc.
See also "realpath".
cat pathReturn the contents of the file named path.
Because, in C, this function returns a char *, there is no way to differentiate between a \0 character in a file and end of string. To handle binary files, use the "read-file" or "download" functions.
checksum csumtype pathThis call computes the MD5, SHAx or CRC checksum of the file named path.
The type of checksum to compute is given by the csumtype parameter which must have one of the following values:
crcCompute the cyclic redundancy check (CRC) specified by POSIX for the cksum command.
md5Compute the MD5 hash (using the md5sum program).
sha1Compute the SHA1 hash (using the sha1sum program).
sha224Compute the SHA224 hash (using the sha224sum program).
sha256Compute the SHA256 hash (using the sha256sum program).
sha384Compute the SHA384 hash (using the sha384sum program).
sha512Compute the SHA512 hash (using the sha512sum program).
The checksum is returned as a printable string.
To get the checksum for a device, use "checksum-device".
To get the checksums for many files, use "checksums-out".
checksum-device csumtype deviceThis call computes the MD5, SHAx or CRC checksum of the contents of the device named device. For the types of checksums supported see the "checksum" command.
checksums-out csumtype directory (sumsfile|-)This command computes the checksums of all regular files in directory and then emits a list of those checksums to the local output file sumsfile.
This can be used for verifying the integrity of a virtual machine. However to be properly secure you should pay attention to the output of the checksum command (it uses the ones from GNU coreutils). In particular when the filename is not printable, coreutils uses a special backslash syntax. For more information, see the GNU coreutils info file.
Use - instead of a filename to read/write from stdin/stdout.
chmod mode pathChange the mode (permissions) of path to mode. Only numeric modes are supported.
Note: When using this command from guestfish, mode by default would be decimal, unless you prefix it with 0 to get octal, ie. use 0700 not 700.
The mode actually set is affected by the umask.
chown owner group pathChange the file owner to owner and group to group.
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).
command 'arguments ...'This call runs a command from the guest filesystem. The filesystem must be mounted, and must contain a compatible operating system (ie. something Linux, with the same or compatible processor architecture).
The single parameter is an argv-style list of arguments. The first element is the name of the program to run. Subsequent elements are parameters. The list must be non-empty (ie. must contain a program name). Note that the command runs directly, and is not invoked via the shell (see "sh").
The return value is anything printed to stdout by the command.
If the command returns a non-zero exit status, then this function returns an error message. The error message string is the content of stderr from the command.
The $PATH environment variable will contain at least /usr/bin and /bin. If you require a program from another location, you should provide the full path in the first parameter.
Shared libraries and data files required by the program must be available on filesystems which are mounted in the correct places. It is the caller's responsibility to ensure all filesystems that are needed are mounted at the right locations.
Because of the message protocol, there is a transfer limit of somewhere between 2MB and 4MB. See "PROTOCOL LIMITS" in guestfs(3).
command-lines 'arguments ...'This is the same as "command", but splits the result into a list of lines.
See also: "sh-lines"
Because of the message protocol, there is a transfer limit of somewhere between 2MB and 4MB. See "PROTOCOL LIMITS" in guestfs(3).
compress-device-out ctype device (zdevice|-) [level:N]This command compresses device and writes it out to the local file zdevice.
The ctype and optional level parameters have the same meaning as in "compress-out".
Use - instead of a filename to read/write from stdin/stdout.
This command has one or more optional arguments. See "OPTIONAL ARGUMENTS".
compress-out ctype file (zfile|-) [level:N]This command compresses file and writes it out to the local file zfile.
The compression program used is controlled by the ctype parameter. Currently this includes: compress, gzip, bzip2, xz or lzop. Some compression types may not be supported by particular builds of libguestfs, in which case you will get an error containing the substring "not supported".
The optional level parameter controls compression level. The meaning and default for this parameter depends on the compression program being used.
Use - instead of a filename to read/write from stdin/stdout.
This command has one or more optional arguments. See "OPTIONAL ARGUMENTS".
config hvparam hvvalueThis can be used to add arbitrary hypervisor parameters of the form -param value. Actually it's not quite arbitrary - we prevent you from setting some parameters which would interfere with parameters that we use.
The first character of hvparam string must be a - (dash).
hvvalue can be NULL.
copy-device-to-device src dest [srcoffset:N] [destoffset:N] [size:N] [sparse:true|false]The four calls "copy-device-to-device", "copy-device-to-file", "copy-file-to-device", and "copy-file-to-file" let you copy from a source (device|file) to a destination (device|file).
Partial copies can be made since you can specify optionally the source offset, destination offset and size to copy. These values are all specified in bytes. If not given, the offsets both default to zero, and the size defaults to copying as much as possible until we hit the end of the source.
The source and destination may be the same object. However overlapping regions may not be copied correctly.
If the destination is a file, it is created if required. If the destination file is not large enough, it is extended.
If the sparse flag is true then the call avoids writing blocks that contain only zeroes, which can help in some situations where the backing disk is thin-provisioned. Note that unless the target is already zeroed, using this option will result in incorrect copying.
This command has one or more optional arguments. See "OPTIONAL ARGUMENTS".
copy-device-to-file src dest [srcoffset:N] [destoffset:N] [size:N] [sparse:true|false]See "copy-device-to-device" for a general overview of this call.
This command has one or more optional arguments. See "OPTIONAL ARGUMENTS".
copy-file-to-device src dest [srcoffset:N] [destoffset:N] [size:N] [sparse:true|false]See "copy-device-to-device" for a general overview of this call.
This command has one or more optional arguments. See "OPTIONAL ARGUMENTS".
copy-file-to-file src dest [srcoffset:N] [destoffset:N] [size:N] [sparse:true|false]See "copy-device-to-device" for a general overview of this call.
This is not the function you want for copying files. This is for copying blocks within existing files. See "cp", "cp-a" and "mv" for general file copying and moving functions.
This command has one or more optional arguments. See "OPTIONAL ARGUMENTS".
copy-size src dest sizeThis command copies exactly size bytes from one source device or file src to another destination device or file dest.
Note this will fail if the source is too short or if the destination is not large enough.
This function is deprecated. In new code, use the "copy-device-to-device" 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.
cp src destThis copies a file from src to dest where dest is either a destination filename or destination directory.
cp-a src destThis copies a file or directory from src to dest recursively using the cp -a command.
cp-r src destThis copies a file or directory from src to dest recursively using the cp -rP command.
Most users should use "cp-a" instead. This command is useful when you don't want to preserve permissions, because the target filesystem does not support it (primarily when writing to DOS FAT filesystems).
dd src destThis command copies from one source device or file src to another destination device or file dest. Normally you would use this to copy to or from a device or partition, for example to duplicate a filesystem.
If the destination is a device, it must be as large or larger than the source file or device, otherwise the copy will fail. This command cannot do partial copies (see "copy-device-to-device").
This function is deprecated. In new code, use the "copy-device-to-device" 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.
device-index deviceThis function takes a device name (eg. "/dev/sdb") and returns the index of the device in the list of devices.
Index numbers start from 0. The named device must exist, for example as a string returned from "list-devices".
See also "list-devices", "part-to-dev".
dfThis command runs the df command to report disk space used.
This command is mostly useful for interactive sessions. It is not intended that you try to parse the output string. Use "statvfs" from programs.
df-hThis command runs the df -h command to report disk space used in human-readable format.
This command is mostly useful for interactive sessions. It is not intended that you try to parse the output string. Use "statvfs" from programs.
disk-format filenameDetect and return the format of the disk image called filename. filename can also be a host device, etc. If the format of the image could not be detected, then "unknown" is returned.
Note that detecting the disk format can be insecure under some circumstances. See "CVE-2010-3851" in guestfs(3).
See also: "DISK IMAGE FORMATS" in guestfs(3)
disk-has-backing-file filenameDetect and return whether the disk image filename has a backing file.
Note that detecting disk features can be insecure under some circumstances. See "CVE-2010-3851" in guestfs(3).
disk-virtual-size filenameDetect and return the virtual size in bytes of the disk image called filename.
Note that detecting disk features can be insecure under some circumstances. See "CVE-2010-3851" in guestfs(3).
dmesgThis returns the kernel messages (dmesg output) from the guest kernel. This is sometimes useful for extended debugging of problems.
Another way to get the same information is to enable verbose messages with "set-verbose" or by setting the environment variable LIBGUESTFS_DEBUG=1 before running the program.
download remotefilename (filename|-)Download file remotefilename and save it as filename on the local machine.
filename can also be a named pipe.
Use - instead of a filename to read/write from stdin/stdout.
download-offset remotefilename (filename|-) offset sizeDownload file remotefilename and save it as filename on the local machine.
remotefilename is read for size bytes starting at offset (this region must be within the file or device).
Note that there is no limit on the amount of data that can be downloaded with this call, unlike with "pread", and this call always reads the full amount unless an error occurs.
See also "download", "pread".
Use - instead of a filename to read/write from stdin/stdout.
drop-caches whattodropThis instructs the guest kernel to drop its page cache, and/or dentries and inode caches. The parameter whattodrop tells the kernel what precisely to drop, see http://linux-mm.org/Drop_Caches
Setting whattodrop to 3 should drop everything.
This automatically calls sync(2) before the operation, so that the maximum guest memory is freed.
du pathThis command runs the du -s command to estimate file space usage for path.
path can be a file or a directory. If path is a directory then the estimate includes the contents of the directory and all subdirectories (recursively).
The result is the estimated size in kilobytes (ie. units of 1024 bytes).
e2fsck device [correct:true|false] [forceall:true|false]This runs the ext2/ext3 filesystem checker on device. It can take the following optional arguments:
correctAutomatically repair the file system. This option will cause e2fsck to automatically fix any filesystem problems that can be safely fixed without human intervention.
This option may not be specified at the same time as the forceall option.
forceallAssume an answer of 'yes' to all questions; allows e2fsck to be used non-interactively.
This option may not be specified at the same time as the correct option.
This command has one or more optional arguments. See "OPTIONAL ARGUMENTS".
e2fsck-f deviceThis runs e2fsck -p -f device, ie. runs the ext2/ext3 filesystem checker on device, noninteractively (-p), even if the filesystem appears to be clean (-f).
This function is deprecated. In new code, use the "e2fsck" 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.
echo-daemon 'words ...'This command concatenates the list of words passed with single spaces between them and returns the resulting string.
You can use this command to test the connection through to the daemon.
See also "ping-daemon".
egrep regex pathThis calls the external egrep program and returns the matching lines.
Because of the message protocol, there is a transfer limit of somewhere between 2MB and 4MB. See "PROTOCOL LIMITS" in guestfs(3).
This function is deprecated. In new code, use the "grep" 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.
egrepi regex pathThis calls the external egrep -i program and returns the matching lines.
Because of the message protocol, there is a transfer limit of somewhere between 2MB and 4MB. See "PROTOCOL LIMITS" in guestfs(3).
This function is deprecated. In new code, use the "grep" 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.
equal file1 file2This compares the two files file1 and file2 and returns true if their content is exactly equal, or false otherwise.
The external cmp(1) program is used for the comparison.
exists pathThis returns true if and only if there is a file, directory (or anything) with the given path name.
See also "is-file", "is-dir", "stat".
extlinux directoryInstall the SYSLINUX bootloader on the device mounted at directory. Unlike "syslinux" which requires a FAT filesystem, this can be used on an ext2/3/4 or btrfs filesystem.
The directory parameter can be either a mountpoint, or a directory within the mountpoint.
You also have to mark the partition as "active" ("part-set-bootable") and a Master Boot Record must be installed (eg. using "pwrite-device") on the first sector of the whole disk. The SYSLINUX package comes with some suitable Master Boot Records. See the extlinux(1) man page for further information.
Additional configuration can be supplied to SYSLINUX by placing a file called extlinux.conf on the filesystem under directory. For further information about the contents of this file, see extlinux(1).
See also "syslinux".
fallocate path lenThis command preallocates a file (containing zero bytes) named path of size len bytes. If the file exists already, it is overwritten.
Do not confuse this with the guestfish-specific alloc command which allocates a file in the host and attaches it as a device.
This function is deprecated. In new code, use the "fallocate64" 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.
fallocate64 path lenThis command preallocates a file (containing zero bytes) named path of size len bytes. If the file exists already, it is overwritten.
Note that this call allocates disk blocks for the file. To create a sparse file use "truncate-size" instead.
The deprecated call "fallocate" does the same, but owing to an oversight it only allowed 30 bit lengths to be specified, effectively limiting the maximum size of files created through that call to 1GB.
Do not confuse this with the guestfish-specific alloc and sparse commands which create a file in the host and attach it as a device.
feature-available 'groups ...'This is the same as "available", but unlike that call it returns a simple true/false boolean result, instead of throwing an exception if a feature is not found. For other documentation see "available".
fgrep pattern pathThis calls the external fgrep program and returns the matching lines.
Because of the message protocol, there is a transfer limit of somewhere between 2MB and 4MB. See "PROTOCOL LIMITS" in guestfs(3).
This function is deprecated. In new code, use the "grep" 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.
fgrepi pattern pathThis calls the external fgrep -i program and returns the matching lines.
Because of the message protocol, there is a transfer limit of somewhere between 2MB and 4MB. See "PROTOCOL LIMITS" in guestfs(3).
This function is deprecated. In new code, use the "grep" 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.
file pathThis call uses the standard file(1) command to determine the type or contents of the file.
This call will also transparently look inside various types of compressed file.
The exact command which runs is file -zb path. Note in particular that the filename is not prepended to the output (the -b option).
The output depends on the output of the underlying file(1) command and it can change in future in ways beyond our control. In other words, the output is not guaranteed by the ABI.
See also: file(1), "vfs-type", "lstat", "is-file", "is-blockdev" (etc), "is-zero".
file-architecture filenameThis detects the architecture of the binary filename, and returns it if known.
Currently defined architectures are:
This string is returned for all 32 bit i386, i486, i586, i686 binaries irrespective of the precise processor requirements of the binary.
64 bit x86-64.
32 bit SPARC.
64 bit SPARC V9 and above.
Intel Itanium.
32 bit Power PC.
64 bit Power PC.
Libguestfs may return other architecture strings in future.
The function works on at least the following types of files:
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
Win32 binaries and DLLs return i386.
Win64 binaries and DLLs return x86_64.
Linux kernel modules
Linux new-style initrd images
some non-x86 Linux vmlinuz kernels
What it can't do currently:
static libraries (libfoo.a)
Linux old-style initrd as compressed ext2 filesystem (RHEL 3)
x86 Linux vmlinuz kernels
x86 vmlinuz images (bzImage format) consist of a mix of 16-, 32- and compressed code, and are horribly hard to unpack. If you want to find the architecture of a kernel, use the architecture of the associated initrd or kernel module(s) instead.
filesize fileThis command returns the size of file in bytes.
To get other stats about a file, use "stat", "lstat", "is-dir", "is-file" etc. To get the size of block devices, use "blockdev-getsize64".
filesystem-available filesystemCheck whether libguestfs supports the named filesystem. The argument filesystem is a filesystem name, such as ext3.
You must call "launch" before using this command.
This is mainly useful as a negative test. If this returns true, it doesn't mean that a particular filesystem can be created or mounted, since filesystems can fail for other reasons such as it being a later version of the filesystem, or having incompatible features, or lacking the right mkfs.<fs> tool.
See also "available", "feature-available", "AVAILABILITY" in guestfs(3).
fill c len pathThis command creates a new file called path. The initial content of the file is len octets of c, where c must be a number in the range [0..255].
To fill a file with zero bytes (sparsely), it is much more efficient to use "truncate-size". To create a file with a pattern of repeating bytes use "fill-pattern".
fill-dir dir nrThis function, useful for testing filesystems, creates nr empty files in the directory dir with names 00000000 through nr-1 (ie. each file name is 8 digits long padded with zeroes).
fill-pattern pattern len pathThis function is like "fill" except that it creates a new file of length len containing the repeating pattern of bytes in pattern. The pattern is truncated if necessary to ensure the length of the file is exactly len bytes.
find directoryThis command lists out all files and directories, recursively, starting at directory. It is essentially equivalent to running the shell command find directory -print but some post-processing happens on the output, described below.
This returns a list of strings without any prefix. Thus if the directory structure was:
/tmp/a
/tmp/b
/tmp/c/dthen the returned list from "find" /tmp would be 4 elements:
a
b
c
c/dIf directory is not a directory, then this command returns an error.
The returned list is sorted.
find0 directory (files|-)This command lists out all files and directories, recursively, starting at directory, placing the resulting list in the external file called files.
This command works the same way as "find" with the following exceptions:
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.
Use - instead of a filename to read/write from stdin/stdout.
findfs-label labelThis command searches the filesystems and returns the one which has the given label. An error is returned if no such filesystem can be found.
To find the label of a filesystem, use "vfs-label".
findfs-uuid uuidThis command searches the filesystems and returns the one which has the given UUID. An error is returned if no such filesystem can be found.
To find the UUID of a filesystem, use "vfs-uuid".
fsck fstype deviceThis runs the filesystem checker (fsck) on device which should have filesystem type fstype.
The returned integer is the status. See fsck(8) for the list of status codes from fsck.
Notes:
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).
This command is entirely equivalent to running fsck -a -t fstype device.
fstrim mountpoint [offset:N] [length:N] [minimumfreeextent:N]Trim the free space in the filesystem mounted on mountpoint. The filesystem must be mounted read-write.
The filesystem contents are not affected, but any free space in the filesystem is "trimmed", that is, given back to the host device, thus making disk images more sparse, allowing unused space in qcow2 files to be reused, etc.
This operation requires support in libguestfs, the mounted filesystem, the host filesystem, qemu and the host kernel. If this support isn't present it may give an error or even appear to run but do nothing.
See also "zero-free-space". That is a slightly different operation that turns free space in the filesystem into zeroes. It is valid to call "fstrim" either instead of, or after calling "zero-free-space".
This command has one or more optional arguments. See "OPTIONAL ARGUMENTS".
get-appendReturn the additional kernel options which are added to the guest kernel command line.
If NULL then no options are added.
get-attach-methodReturn the current backend.
See "set-backend" and "BACKEND" in guestfs(3).
This function is deprecated. In new code, use the "get-backend" 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.
get-autosyncGet the autosync flag.
get-backendReturn the current backend.
This handle property was previously called the "attach method".
See "set-backend" and "BACKEND" in guestfs(3).
get-cachedirGet the directory used by the handle to store the appliance cache.
get-directReturn the direct appliance mode flag.
get-e2attrs fileThis returns the file attributes associated with file.
The attributes are a set of bits associated with each inode which affect the behaviour of the file. The attributes are returned as a string of letters (described below). The string may be empty, indicating that no file attributes are set for this file.
These attributes are only present when the file is located on an ext2/3/4 filesystem. Using this call on other filesystem types will result in an error.
The characters (file attributes) in the returned string are currently:
When the file is accessed, its atime is not modified.
The file is append-only.
The file is compressed on-disk.
(Directories only.) Changes to this directory are written synchronously to disk.
The file is not a candidate for backup (see dump(8)).
The file has compression errors.
The file is using extents.
The file is storing its blocks in units of the filesystem blocksize instead of sectors.
(Directories only.) The directory is using hashed trees.
The file is immutable. It cannot be modified, deleted or renamed. No link can be created to this file.
The file is data-journaled.
When the file is deleted, all its blocks will be zeroed.
Changes to this file are written synchronously to disk.
(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.
For a file, this disables tail-merging. (Not used by upstream implementations of ext2.)
When the file is deleted, its blocks will be saved, allowing the file to be undeleted.
The raw contents of the compressed file may be accessed.
The compressed file is dirty.
More file attributes may be added to this list later. Not all file attributes may be set for all kinds of files. For detailed information, consult the chattr(1) man page.
See also "set-e2attrs".
Don't confuse these attributes with extended attributes (see "getxattr").
get-e2generation fileThis returns the ext2 file generation of a file. The generation (which used to be called the "version") is a number associated with an inode. This is most commonly used by NFS servers.
The generation is only present when the file is located on an ext2/3/4 filesystem. Using this call on other filesystem types will result in an error.
See "set-e2generation".
get-e2label deviceThis returns the ext2/3/4 filesystem label of the filesystem on device.
This function is deprecated. In new code, use the "vfs-label" 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.
get-e2uuid deviceThis returns the ext2/3/4 filesystem UUID of the filesystem on device.
This function is deprecated. In new code, use the "vfs-uuid" 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.
get-hvReturn the current hypervisor binary.
This is always non-NULL. If it wasn't set already, then this will return the default qemu binary name.
get-libvirt-requested-credential-challenge indexGet the challenge (provided by libvirt) for the index'th requested credential. If libvirt did not provide a challenge, this returns the empty string "".
See "LIBVIRT AUTHENTICATION" in guestfs(3) for documentation and example code.
get-libvirt-requested-credential-defresult indexGet the default result (provided by libvirt) for the index'th requested credential. If libvirt did not provide a default result, this returns the empty string "".
See "LIBVIRT AUTHENTICATION" in guestfs(3) for documentation and example code.
get-libvirt-requested-credential-prompt indexGet the prompt (provided by libvirt) for the index'th requested credential. If libvirt did not provide a prompt, this returns the empty string "".
See "LIBVIRT AUTHENTICATION" in guestfs(3) for documentation and example code.
get-libvirt-requested-credentialsThis should only be called during the event callback for events of type GUESTFS_EVENT_LIBVIRT_AUTH.
Return the list of credentials requested by libvirt. Possible values are a subset of the strings provided when you called "set-libvirt-supported-credentials".
See "LIBVIRT AUTHENTICATION" in guestfs(3) for documentation and example code.
get-memsizeThis gets the memory size in megabytes allocated to the hypervisor.
If "set-memsize" was not called on this handle, and if LIBGUESTFS_MEMSIZE was not set, then this returns the compiled-in default value for memsize.
For more information on the architecture of libguestfs, see guestfs(3).
get-networkThis returns the enable network flag.
get-pathReturn the current search path.
This is always non-NULL. If it wasn't set already, then this will return the default path.
get-pgroupThis returns the process group flag.
get-pidReturn the process ID of the hypervisor. If there is no hypervisor running, then this will return an error.
This is an internal call used for debugging and testing.
get-programGet the program name. See "set-program".
get-qemuReturn the current hypervisor binary (usually qemu).
This is always non-NULL. If it wasn't set already, then this will return the default qemu binary name.
This function is deprecated. In new code, use the "get-hv" 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.
get-recovery-procReturn the recovery process enabled flag.
get-selinuxThis returns the current setting of the selinux flag which is passed to the appliance at boot time. See "set-selinux".
For more information on the architecture of libguestfs, see guestfs(3).
get-smpThis returns the number of virtual CPUs assigned to the appliance.
get-tmpdirGet the directory used by the handle to store temporary files.
get-traceReturn the command trace flag.
get-umaskReturn the current umask. By default the umask is 022 unless it has been set by calling "umask".
get-verboseThis returns the verbose messages flag.
getconThis gets the SELinux security context of the daemon.
See the documentation about SELINUX in guestfs(3), and "setcon"
getxattr path nameGet a single extended attribute from file path named name. This call follows symlinks. If you want to lookup an extended attribute for the symlink itself, use "lgetxattr".
Normally it is better to get all extended attributes from a file in one go by calling "getxattrs". However some Linux filesystem implementations are buggy and do not provide a way to list out attributes. For these filesystems (notably ntfs-3g) you have to know the names of the extended attributes you want in advance and call this function.
Extended attribute values are blobs of binary data. If there is no extended attribute named name, this returns an error.
See also: "getxattrs", "lgetxattr", attr(5).
getxattrs pathThis call lists the extended attributes of the file or directory path.
At the system call level, this is a combination of the listxattr(2) and getxattr(2) calls.
See also: "lgetxattrs", attr(5).
glob-expand patternThis command searches for all the pathnames matching pattern according to the wildcard expansion rules used by the shell.
If no paths match, then this returns an empty list (note: not an error).
It is just a wrapper around the C glob(3) function with flags GLOB_MARK|GLOB_BRACE. See that manual page for more details.
Notice that there is no equivalent command for expanding a device name (eg. /dev/sd*). Use "list-devices", "list-partitions" etc functions instead.
grep regex path [extended:true|false] [fixed:true|false] [insensitive:true|false] [compressed:true|false]This calls the external grep program and returns the matching lines.
The optional flags are:
extendedUse extended regular expressions. This is the same as using the -E flag.
fixedMatch fixed (don't use regular expressions). This is the same as using the -F flag.
insensitiveMatch case-insensitive. This is the same as using the -i flag.
compressedUse zgrep instead of grep. This allows the input to be compress- or gzip-compressed.
This command has one or more optional arguments. See "OPTIONAL ARGUMENTS".
Because of the message protocol, there is a transfer limit of somewhere between 2MB and 4MB. See "PROTOCOL LIMITS" in guestfs(3).
grepi regex pathThis calls the external grep -i program and returns the matching lines.
Because of the message protocol, there is a transfer limit of somewhere between 2MB and 4MB. See "PROTOCOL LIMITS" in guestfs(3).
This function is deprecated. In new code, use the "grep" 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.
grub-install root deviceThis command installs GRUB 1 (the Grand Unified Bootloader) on device, with the root directory being root.
Notes:
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 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:
(hd0) /dev/vdareplacing /dev/vda with the name of the installation device.
head pathThis command returns up to the first 10 lines of a file as a list of strings.
Because of the message protocol, there is a transfer limit of somewhere between 2MB and 4MB. See "PROTOCOL LIMITS" in guestfs(3).
head-n nrlines pathIf the parameter nrlines is a positive number, this returns the first nrlines lines of the file path.
If the parameter nrlines is a negative number, this returns lines from the file path, excluding the last nrlines lines.
If the parameter nrlines is zero, this returns an empty list.
Because of the message protocol, there is a transfer limit of somewhere between 2MB and 4MB. See "PROTOCOL LIMITS" in guestfs(3).
hexdump pathThis runs hexdump -C on the given path. The result is the human-readable, canonical hex dump of the file.
Because of the message protocol, there is a transfer limit of somewhere between 2MB and 4MB. See "PROTOCOL LIMITS" in guestfs(3).
hivex-closeClose the current hivex handle.
This is a wrapper around the hivex(3) call of the same name.
hivex-commit filenameCommit (write) changes to the hive.
If the optional filename parameter is null, then the changes are written back to the same hive that was opened. If this is not null then they are written to the alternate filename given and the original hive is left untouched.
This is a wrapper around the hivex(3) call of the same name.
hivex-node-add-child parent nameAdd a child node to parent named name.
This is a wrapper around the hivex(3) call of the same name.
hivex-node-children nodehReturn the list of nodes which are subkeys of nodeh.
This is a wrapper around the hivex(3) call of the same name.
hivex-node-delete-child nodehDelete nodeh, recursively if necessary.
This is a wrapper around the hivex(3) call of the same name.
hivex-node-get-child nodeh nameReturn the child of nodeh with the name name, if it exists. This can return 0 meaning the name was not found.
This is a wrapper around the hivex(3) call of the same name.
hivex-node-get-value nodeh keyReturn the value attached to nodeh which has the name key, if it exists. This can return 0 meaning the key was not found.
This is a wrapper around the hivex(3) call of the same name.
hivex-node-name nodehReturn the name of nodeh.
This is a wrapper around the hivex(3) call of the same name.
hivex-node-parent nodehReturn the parent node of nodeh.
This is a wrapper around the hivex(3) call of the same name.
hivex-node-set-value nodeh key t valSet or replace a single value under the node nodeh. The key is the name, t is the type, and val is the data.
This is a wrapper around the hivex(3) call of the same name.
hivex-node-values nodehReturn the array of (key, datatype, data) tuples attached to nodeh.
This is a wrapper around the hivex(3) call of the same name.
hivex-open filename [verbose:true|false] [debug:true|false] [write:true|false]Open the Windows Registry hive file named filename. If there was any previous hivex handle associated with this guestfs session, then it is closed.
This is a wrapper around the hivex(3) call of the same name.
This command has one or more optional arguments. See "OPTIONAL ARGUMENTS".
hivex-rootReturn the root node of the hive.
This is a wrapper around the hivex(3) call of the same name.
hivex-value-key valuehReturn the key (name) field of a (key, datatype, data) tuple.
This is a wrapper around the hivex(3) call of the same name.
hivex-value-type valuehReturn the data type field from a (key, datatype, data) tuple.
This is a wrapper around the hivex(3) call of the same name.
hivex-value-utf8 valuehThis calls "hivex-value-value" (which returns the data field from a hivex value tuple). It then assumes that the field is a UTF-16LE string and converts the result to UTF-8 (or if this is not possible, it returns an error).
This is useful for reading strings out of the Windows registry. However it is not foolproof because the registry is not strongly-typed and fields can contain arbitrary or unexpected data.
hivex-value-value valuehReturn the data field of a (key, datatype, data) tuple.
This is a wrapper around the hivex(3) call of the same name.
See also: "hivex-value-utf8".
initrd-cat initrdpath filenameThis command unpacks the file filename from the initrd file called initrdpath. The filename must be given without the initial / character.
For example, in guestfish you could use the following command to examine the boot script (usually called /init) contained in a Linux initrd or initramfs image:
initrd-cat /boot/initrd-<version>.img initSee also "initrd-list".
Because of the message protocol, there is a transfer limit of somewhere between 2MB and 4MB. See "PROTOCOL LIMITS" in guestfs(3).
initrd-list pathThis command lists out files contained in an initrd.
The files are listed without any initial / character. The files are listed in the order they appear (not necessarily alphabetical). Directory names are listed as separate items.
Old Linux kernels (2.4 and earlier) used a compressed ext2 filesystem as initrd. We only support the newer initramfs format (compressed cpio files).
inotify-add-watch path maskWatch path for the events listed in mask.
Note that if path is a directory then events within that directory are watched, but this does not happen recursively (in subdirectories).
Note for non-C or non-Linux callers: the inotify events are defined by the Linux kernel ABI and are listed in /usr/include/sys/inotify.h.
inotify-closeThis closes the inotify handle which was previously opened by inotify_init. It removes all watches, throws away any pending events, and deallocates all resources.
inotify-filesThis function is a helpful wrapper around "inotify-read" which just returns a list of pathnames of objects that were touched. The returned pathnames are sorted and deduplicated.
inotify-init maxeventsThis command creates a new inotify handle. The inotify subsystem can be used to notify events which happen to objects in the guest filesystem.
maxevents is the maximum number of events which will be queued up between calls to "inotify-read" or "inotify-files". If this is passed as 0, then the kernel (or previously set) default is used. For Linux 2.6.29 the default was 16384 events. Beyond this limit, the kernel throws away events, but records the fact that it threw them away by setting a flag IN_Q_OVERFLOW in the returned structure list (see "inotify-read").
Before any events are generated, you have to add some watches to the internal watch list. See: "inotify-add-watch" and "inotify-rm-watch".
Queued up events should be read periodically by calling "inotify-read" (or "inotify-files" which is just a helpful wrapper around "inotify-read"). If you don't read the events out often enough then you risk the internal queue overflowing.
The handle should be closed after use by calling "inotify-close". This also removes any watches automatically.
See also inotify(7) for an overview of the inotify interface as exposed by the Linux kernel, which is roughly what we expose via libguestfs. Note that there is one global inotify handle per libguestfs instance.
inotify-readReturn the complete queue of events that have happened since the previous read call.
If no events have happened, this returns an empty list.
Note: In order to make sure that all events have been read, you must call this function repeatedly until it returns an empty list. The reason is that the call will read events up to the maximum appliance-to-host message size and leave remaining events in the queue.
inotify-rm-watch wdRemove a previously defined inotify watch. See "inotify-add-watch".
inspect-get-arch rootThis returns the architecture of the inspected operating system. The possible return values are listed under "file-architecture".
If the architecture could not be determined, then the string unknown is returned.
Please read "INSPECTION" in guestfs(3) for more details.
inspect-get-distro rootThis returns the distro (distribution) of the inspected operating system.
Currently defined distros are:
Arch Linux.
Buildroot-derived distro, but not one we specifically recognize.
CentOS.
Cirros.
Debian.
Fedora.
FreeDOS.
Gentoo.
Linux Mint.
Mageia.
Mandriva.
MeeGo.
OpenBSD.
OpenSUSE.
Pardus.
Some Red Hat-derived distro.
Red Hat Enterprise Linux.
Scientific Linux.
Slackware.
SuSE Linux Enterprise Server or Desktop.
Some openSuSE-derived distro.
ttylinux.
Ubuntu.
The distro could not be determined.
Windows does not have distributions. This string is returned if the OS type is Windows.
Future versions of libguestfs may return other strings here. The caller should be prepared to handle any string.
Please read "INSPECTION" in guestfs(3) for more details.
inspect-get-drive-mappings rootThis call is useful for Windows which uses a primitive system of assigning drive letters (like "C:") to partitions. This inspection API examines the Windows Registry to find out how disks/partitions are mapped to drive letters, and returns a hash table as in the example below:
C => /dev/vda2
E => /dev/vdb1
F => /dev/vdc1Note that keys are drive letters. For Windows, the key is case insensitive and just contains the drive letter, without the customary colon separator character.
In future we may support other operating systems that also used drive letters, but the keys for those might not be case insensitive and might be longer than 1 character. For example in OS-9, hard drives were named h0, h1 etc.
For Windows guests, currently only hard drive mappings are returned. Removable disks (eg. DVD-ROMs) are ignored.
For guests that do not use drive mappings, or if the drive mappings could not be determined, this returns an empty hash table.
Please read "INSPECTION" in guestfs(3) for more details. See also "inspect-get-mountpoints", "inspect-get-filesystems".
inspect-get-filesystems rootThis returns a list of all the filesystems that we think are associated with this operating system. This includes the root filesystem, other ordinary filesystems, and non-mounted devices like swap partitions.
In the case of a multi-boot virtual machine, it is possible for a filesystem to be shared between operating systems.
Please read "INSPECTION" in guestfs(3) for more details. See also "inspect-get-mountpoints".
inspect-get-format rootThis returns the format of the inspected operating system. You can use it to detect install images, live CDs and similar.
Currently defined formats are:
This is an installed operating system.
The disk image being inspected is not an installed operating system, but a bootable install disk, live CD, or similar.
The format of this disk image is not known.
Future versions of libguestfs may return other strings here. The caller should be prepared to handle any string.
Please read "INSPECTION" in guestfs(3) for more details.
inspect-get-hostname rootThis function returns the hostname of the operating system as found by inspection of the guest's configuration files.
If the hostname could not be determined, then the string unknown is returned.
Please read "INSPECTION" in guestfs(3) for more details.
inspect-get-icon root [favicon:true|false] [highquality:true|false]This function returns an icon corresponding to the inspected operating system. The icon is returned as a buffer containing a PNG image (re-encoded to PNG if necessary).
If it was not possible to get an icon this function returns a zero-length (non-NULL) buffer. Callers must check for this case.
Libguestfs will start by looking for a file called /etc/favicon.png or C:\etc\favicon.png and if it has the correct format, the contents of this file will be returned. You can disable favicons by passing the optional favicon boolean as false (default is true).
If finding the favicon fails, then we look in other places in the guest for a suitable icon.
If the optional highquality boolean is true then only high quality icons are returned, which means only icons of high resolution with an alpha channel. The default (false) is to return any icon we can, even if it is of substandard quality.
Notes:
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 program from the icoutils package, and several programs (bmptopnm, pnmtopng, pamcut) from the netpbm package. These must be installed separately.
Operating system icons are usually trademarks. Seek legal advice before using trademarks in applications.
This command has one or more optional arguments. See "OPTIONAL ARGUMENTS".
inspect-get-major-version rootThis returns the major version number of the inspected operating system.
Windows uses a consistent versioning scheme which is not reflected in the popular public names used by the operating system. Notably the operating system known as "Windows 7" is really version 6.1 (ie. major = 6, minor = 1). You can find out the real versions corresponding to releases of Windows by consulting Wikipedia or MSDN.
If the version could not be determined, then 0 is returned.
Please read "INSPECTION" in guestfs(3) for more details.
inspect-get-minor-version rootThis returns the minor version number of the inspected operating system.
If the version could not be determined, then 0 is returned.
Please read "INSPECTION" in guestfs(3) for more details. See also "inspect-get-major-version".
inspect-get-mountpoints rootThis returns a hash of where we think the filesystems associated with this operating system should be mounted. Callers should note that this is at best an educated guess made by reading configuration files such as /etc/fstab. In particular note that this may return filesystems which are non-existent or not mountable and callers should be prepared to handle or ignore failures if they try to mount them.
Each element in the returned hashtable has a key which is the path of the mountpoint (eg. /boot) and a value which is the filesystem that would be mounted there (eg. /dev/sda1).
Non-mounted devices such as swap devices are not returned in this list.
For operating systems like Windows which still use drive letters, this call will only return an entry for the first drive "mounted on" /. For information about the mapping of drive letters to partitions, see "inspect-get-drive-mappings".
Please read "INSPECTION" in guestfs(3) for more details. See also "inspect-get-filesystems".
inspect-get-package-format rootThis function and "inspect-get-package-management" return the package format and package management tool used by the inspected operating system. For example for Fedora these functions would return rpm (package format) and yum (package management).
This returns the string unknown if we could not determine the package format or if the operating system does not have a real packaging system (eg. Windows).
Possible strings include: rpm, deb, ebuild, pisi, pacman, pkgsrc. Future versions of libguestfs may return other strings.
Please read "INSPECTION" in guestfs(3) for more details.
inspect-get-package-management root"inspect-get-package-format" and this function return the package format and package management tool used by the inspected operating system. For example for Fedora these functions would return rpm (package format) and yum (package management).
This returns the string unknown if we could not determine the package management tool or if the operating system does not have a real packaging system (eg. Windows).
Possible strings include: yum, up2date, apt (for all Debian derivatives), portage, pisi, pacman, urpmi, zypper. Future versions of libguestfs may return other strings.
Please read "INSPECTION" in guestfs(3) for more details.
inspect-get-product-name rootThis returns the product name of the inspected operating system. The product name is generally some freeform string which can be displayed to the user, but should not be parsed by programs.
If the product name could not be determined, then the string unknown is returned.
Please read "INSPECTION" in guestfs(3) for more details.
inspect-get-product-variant rootThis returns the product variant of the inspected operating system.
For Windows guests, this returns the contents of the Registry key HKLM\Software\Microsoft\Windows NT\CurrentVersion InstallationType which is usually a string such as Client or Server (other values are possible). This can be used to distinguish consumer and enterprise versions of Windows that have the same version number (for example, Windows 7 and Windows 2008 Server are both version 6.1, but the former is Client and the latter is Server).
For enterprise Linux guests, in future we intend this to return the product variant such as Desktop, Server and so on. But this is not implemented at present.
If the product variant could not be determined, then the string unknown is returned.
Please read "INSPECTION" in guestfs(3) for more details. See also "inspect-get-product-name", "inspect-get-major-version".
inspect-get-rootsThis function is a convenient way to get the list of root devices, as returned from a previous call to "inspect-os", but without redoing the whole inspection process.
This returns an empty list if either no root devices were found or the caller has not called "inspect-os".
Please read "INSPECTION" in guestfs(3) for more details.
inspect-get-type rootThis returns the type of the inspected operating system. Currently defined types are:
Any Linux-based operating system.
Any Microsoft Windows operating system.
FreeBSD.
NetBSD.
OpenBSD.
GNU/Hurd.
MS-DOS, FreeDOS and others.
The operating system type could not be determined.
Future versions of libguestfs may return other strings here. The caller should be prepared to handle any string.
Please read "INSPECTION" in guestfs(3) for more details.
inspect-get-windows-current-control-set rootThis returns the Windows CurrentControlSet of the inspected guest. The CurrentControlSet is a registry key name such as ControlSet001.
This call assumes that the guest is Windows and that the Registry could be examined by inspection. If this is not the case then an error is returned.
Please read "INSPECTION" in guestfs(3) for more details.
inspect-get-windows-systemroot rootThis returns the Windows systemroot of the inspected guest. The systemroot is a directory path such as /WINDOWS.
This call assumes that the guest is Windows and that the systemroot could be determined by inspection. If this is not the case then an error is returned.
Please read "INSPECTION" in guestfs(3) for more details.
inspect-is-live rootIf "inspect-get-format" returns installer (this is an install disk), then this returns true if a live image was detected on the disk.
Please read "INSPECTION" in guestfs(3) for more details.
inspect-is-multipart rootIf "inspect-get-format" returns installer (this is an install disk), then this returns true if the disk is part of a set.
Please read "INSPECTION" in guestfs(3) for more details.
inspect-is-netinst rootIf "inspect-get-format" returns installer (this is an install disk), then this returns true if the disk is a network installer, ie. not a self-contained install CD but one which is likely to require network access to complete the install.
Please read "INSPECTION" in guestfs(3) for more details.
inspect-list-applications rootReturn the list of applications installed in the operating system.
Note: This call works differently from other parts of the inspection API. You have to call "inspect-os", then "inspect-get-mountpoints", then mount up the disks, before calling this. Listing applications is a significantly more difficult operation which requires access to the full filesystem. Also note that unlike the other "inspect-get-*" calls which are just returning data cached in the libguestfs handle, this call actually reads parts of the mounted filesystems during the call.
This returns an empty list if the inspection code was not able to determine the list of applications.
The application structure contains the following fields:
app_nameThe name of the application. For Red Hat-derived and Debian-derived Linux guests, this is the package name.
app_display_nameThe display name of the application, sometimes localized to the install language of the guest operating system.
If unavailable this is returned as an empty string "". Callers needing to display something can use app_name instead.
app_epochFor package managers which use epochs, this contains the epoch of the package (an integer). If unavailable, this is returned as 0.
app_versionThe version string of the application or package. If unavailable this is returned as an empty string "".
app_releaseThe release string of the application or package, for package managers that use this. If unavailable this is returned as an empty string "".
app_install_pathThe installation path of the application (on operating systems such as Windows which use installation paths). This path is in the format used by the guest operating system, it is not a libguestfs path.
If unavailable this is returned as an empty string "".
app_trans_pathThe install path translated into a libguestfs path. If unavailable this is returned as an empty string "".
app_publisherThe name of the publisher of the application, for package managers that use this. If unavailable this is returned as an empty string "".
app_urlThe URL (eg. upstream URL) of the application. If unavailable this is returned as an empty string "".
app_source_packageFor packaging systems which support this, the name of the source package. If unavailable this is returned as an empty string "".
app_summaryA short (usually one line) description of the application or package. If unavailable this is returned as an empty string "".
app_descriptionA longer description of the application or package. If unavailable this is returned as an empty string "".
Please read "INSPECTION" in guestfs(3) for more details.
This function is deprecated. In new code, use the "inspect-list-applications2" 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.
inspect-list-applications2 rootReturn the list of applications installed in the operating system.
Note: This call works differently from other parts of the inspection API. You have to call "inspect-os", then "inspect-get-mountpoints", then mount up the disks, before calling this. Listing applications is a significantly more difficult operation which requires access to the full filesystem. Also note that unlike the other "inspect-get-*" calls which are just returning data cached in the libguestfs handle, this call actually reads parts of the mounted filesystems during the call.
This returns an empty list if the inspection code was not able to determine the list of applications.
The application structure contains the following fields:
app2_nameThe name of the application. For Red Hat-derived and Debian-derived Linux guests, this is the package name.
app2_display_nameThe display name of the application, sometimes localized to the install language of the guest operating system.
If unavailable this is returned as an empty string "". Callers needing to display something can use app2_name instead.
app2_epochFor package managers which use epochs, this contains the epoch of the package (an integer). If unavailable, this is returned as 0.
app2_versionThe version string of the application or package. If unavailable this is returned as an empty string "".
app2_releaseThe release string of the application or package, for package managers that use this. If unavailable this is returned as an empty string "".
app2_archThe architecture string of the application or package, for package managers that use this. If unavailable this is returned as an empty string "".
app2_install_pathThe installation path of the application (on operating systems such as Windows which use installation paths). This path is in the format used by the guest operating system, it is not a libguestfs path.
If unavailable this is returned as an empty string "".
app2_trans_pathThe install path translated into a libguestfs path. If unavailable this is returned as an empty string "".
app2_publisherThe name of the publisher of the application, for package managers that use this. If unavailable this is returned as an empty string "".
app2_urlThe URL (eg. upstream URL) of the application. If unavailable this is returned as an empty string "".
app2_source_packageFor packaging systems which support this, the name of the source package. If unavailable this is returned as an empty string "".
app2_summaryA short (usually one line) description of the application or package. If unavailable this is returned as an empty string "".
app2_descriptionA longer description of the application or package. If unavailable this is returned as an empty string "".
Please read "INSPECTION" in guestfs(3) for more details.
inspect-osThis function uses other libguestfs functions and certain heuristics to inspect the disk(s) (usually disks belonging to a virtual machine), looking for operating systems.
The list returned is empty if no operating systems were found.
If one operating system was found, then this returns a list with a single element, which is the name of the root filesystem of this operating system. It is also possible for this function to return a list containing more than one element, indicating a dual-boot or multi-boot virtual machine, with each element being the root filesystem of one of the operating systems.
You can pass the root string(s) returned to other "inspect-get-*" functions in order to query further information about each operating system, such as the name and version.
This function uses other libguestfs features such as "mount-ro" and "umount-all" in order to mount and unmount filesystems and look at the contents. This should be called with no disks currently mounted. The function may also use Augeas, so any existing Augeas handle will be closed.
This function cannot decrypt encrypted disks. The caller must do that first (supplying the necessary keys) if the disk is encrypted.
Please read "INSPECTION" in guestfs(3) for more details.
See also "list-filesystems".
is-blockdev path [followsymlinks:true|false]This returns true if and only if there is a block device with the given path name.
If the optional flag followsymlinks is true, then a symlink (or chain of symlinks) that ends with a block device also causes the function to return true.
See also "stat".
This command has one or more optional arguments. See "OPTIONAL ARGUMENTS".
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-configThis returns true iff this handle is being configured (in the CONFIG state).
For more information on states, see guestfs(3).
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 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 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 deviceThis command tests whether device is a logical volume, and returns true iff this is the case.
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 pathThis returns true if and only if there is a symbolic link with the given path name.
See also "stat".
is-whole-device deviceThis returns true if and only if device refers to a whole block device. That is, not a partition or a logical device.
is-zero pathThis returns true iff the file exists and the file is empty or it contains all zero bytes.
is-zero-device deviceThis 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 isofileThis 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 devicedevice 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
journal-closeClose the journal handle.
journal-getRead 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.
journal-get-data-thresholdGet 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.
See also "journal-set-data-threshold".
journal-nextMove to the next journal entry. You have to call this at least once after opening the handle before you are able to read data.
The returned boolean tells you if there are any more journal records to read. true means you can read the next record (eg. using "journal-get-data"), and false means you have reached the end of the journal.
journal-open directoryOpen the systemd journal located in directory. Any previously opened journal handle is closed.
The contents of the journal can be read using "journal-next" and "journal-get".
After you have finished using the journal, you should close the handle by calling "journal-close".
journal-set-data-threshold thresholdSet the 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 you set this to 0, then the threshold is unlimited.
See also "journal-get-data-threshold".
journal-skip skipSkip forwards (skip ≥ 0) or backwards (skip < 0) in the journal.
The number of entries actually skipped is returned (note rskip ≥ 0). If this is not the same as the absolute value of the skip parameter (|skip|) you passed in then it means you have reached the end or the start of the journal.
kill-subprocessThis kills the hypervisor.
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.
launchYou 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 owner group pathChange 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-allThis 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 diskgroupReturn 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 diskgroupReturn 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 diskgroupReturn 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-allThis is essentially the opposite of "ldmtool-create-all". It removes the device mapper mappings for all Windows dynamic disk volumes
ldmtool-scanThis 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 '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 diskgroup volumeReturn 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 diskgroup volumeReturn the list of partitions in the volume named volume in the disk group with GUID diskgroup.
ldmtool-volume-type diskgroup volumeReturn 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 path nameGet a single extended attribute from file path named name. If path is a symlink, then this call returns an extended attribute from the symlink.
Normally it is better to get all extended attributes from a file in one go by calling "getxattrs". However some Linux filesystem implementations are buggy and do not provide a way to list out attributes. For these filesystems (notably ntfs-3g) you have to know the names of the extended attributes you want in advance and call this function.
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 pathThis is the same as "getxattrs", but if path is a symbolic link, then it returns the extended attributes of the link itself.
list-9pList all 9p filesystems attached to the guest. A list of mount tags is returned.
list-devicesList all the block devices.
The full block device names are returned, eg. /dev/sda.
See also "list-filesystems".
list-disk-labelsIf you add drives using the optional label parameter of "add-drive-opts", you can use this call to map between disk labels, and raw block device and partition names (like /dev/sda and /dev/sda1).
This returns a hashtable, where keys are the disk labels (without the /dev/disk/guestfs prefix), and the values are the full raw block device and partition names (eg. /dev/sda and /dev/sda1).
list-dm-devicesList all device mapper devices.
The returned list contains /dev/mapper/* devices, eg. ones created by a previous call to "luks-open".
Device mapper devices which correspond to logical volumes are not returned in this list. Call "lvs" if you want to list logical volumes.
list-filesystemsThis inspection command looks for filesystems on partitions, block devices and logical volumes, returning a list of mountables containing filesystems and their type.
The return value is a hash, where the keys are the devices containing filesystems, and the values are the filesystem types. For example:
"/dev/sda1" => "ntfs"
"/dev/sda2" => "ext2"
"/dev/vg_guest/lv_root" => "ext4"
"/dev/vg_guest/lv_swap" => "swap"The key is not necessarily a block device. It may also be an opaque 'mountable' string which can be passed to "mount".
The value can have the special value "unknown", meaning the content of the device is undetermined or empty. "swap" means a Linux swap partition.
This command runs other libguestfs commands, which might include "mount" and "umount", and therefore you should use this soon after launch and only when nothing is mounted.
Not all of the filesystems returned will be mountable. In particular, swap partitions are returned in the list. Also this command does not check that each filesystem found is valid and mountable, and some filesystems might be mountable but require special options. Filesystems may not all belong to a single logical operating system (use "inspect-os" to look for OSes).
list-ldm-partitionsThis function returns all Windows dynamic disk partitions that were found at launch time. It returns a list of device names.
list-ldm-volumesThis function returns all Windows dynamic disk volumes that were found at launch time. It returns a list of device names.
list-md-devicesList all Linux md devices.
list-partitionsList all the partitions detected on all block devices.
The full partition device names are returned, eg. /dev/sda1
This does not return logical volumes. For that you will need to call "lvs".
See also "list-filesystems".
ll directoryList the files in directory (relative to the root directory, there is no cwd) in the format of 'ls -la'.
This command is mostly useful for interactive sessions. It is not intended that you try to parse the output string.
llz directoryList the files in directory in the format of 'ls -laZ'.
This command is mostly useful for interactive sessions. It is not intended that you try to parse the output string.
ln target linknameThis command creates a hard link using the ln command.
ln-f target linknameThis command creates a hard link using the ln -f command. The -f option removes the link (linkname) if it exists already.
ln-s target linknameThis command creates a symbolic link using the ln -s command.
ln-sf target linknameThis command creates a symbolic link using the ln -sf command, The -f option removes the link (linkname) if it exists already.
lremovexattr xattr pathThis is the same as "removexattr", but if path is a symbolic link, then it removes an extended attribute of the link itself.
ls directoryList the files in directory (relative to the root directory, there is no cwd). The '.' and '..' entries are not returned, but hidden files are shown.
ls0 dir (filenames|-)This specialized command is used to get a listing of the filenames in the directory dir. The list of filenames is written to the local file filenames (on the host).
In the output file, the filenames are separated by \0 characters.
. and .. are not returned. The filenames are not sorted.
Use - instead of a filename to read/write from stdin/stdout.
lsetxattr xattr val vallen pathThis is the same as "setxattr", but if path is a symbolic link, then it sets an extended attribute of the link itself.
lstat pathReturns file information for the given path.
This is the same as "stat" except that if path is a symbolic link, then the link is stat-ed, not the file it refers to.
This is the same as the lstat(2) system call.
lstatlist path 'names ...'This call allows you to perform the "lstat" operation on multiple files, where all files are in the directory path. names is the list of files from this directory.
On return you get a list of stat structs, with a one-to-one correspondence to the names list. If any name did not exist or could not be lstat'd, then the ino field of that structure is set to -1.
This call is intended for programs that want to efficiently list a directory contents without making many round-trips. See also "lxattrlist" for a similarly efficient call for getting extended attributes.
luks-add-key device keyslotThis command adds a new key on LUKS device device. key is any existing key, and is used to access the device. newkey is the new key to add. keyslot is the key slot that will be replaced.
Note that if keyslot already contains a key, then this command will fail. You have to use "luks-kill-slot" first to remove that key.
This command has one or more key or passphrase parameters. Guestfish will prompt for these separately.
luks-close deviceThis closes a LUKS device that was created earlier by "luks-open" or "luks-open-ro". The device parameter must be the name of the LUKS mapping device (ie. /dev/mapper/mapname) and not the name of the underlying block device.
luks-format device keyslotThis command erases existing data on device and formats the device as a LUKS encrypted device. key is the initial key, which is added to key slot slot. (LUKS supports 8 key slots, numbered 0-7).
This command has one or more key or passphrase parameters. Guestfish will prompt for these separately.
luks-format-cipher device keyslot cipherThis command is the same as "luks-format" but it also allows you to set the cipher used.
This command has one or more key or passphrase parameters. Guestfish will prompt for these separately.
luks-kill-slot device keyslotThis command deletes the key in key slot keyslot from the encrypted LUKS device device. key must be one of the other keys.
This command has one or more key or passphrase parameters. Guestfish will prompt for these separately.
luks-open device mapnameThis command opens a block device which has been encrypted according to the Linux Unified Key Setup (LUKS) standard.
device is the encrypted block device or partition.
The caller must supply one of the keys associated with the LUKS block device, in the key parameter.
This creates a new block device called /dev/mapper/mapname. Reads and writes to this block device are decrypted from and encrypted to the underlying device respectively.
If this block device contains LVM volume groups, then calling "vgscan" followed by "vg-activate-all" will make them visible.
Use "list-dm-devices" to list all device mapper devices.
This command has one or more key or passphrase parameters. Guestfish will prompt for these separately.
luks-open-ro device mapnameThis is the same as "luks-open" except that a read-only mapping is created.
This command has one or more key or passphrase parameters. Guestfish will prompt for these separately.
lvcreate logvol volgroup mbytesThis creates an LVM logical volume called logvol on the volume group volgroup, with size megabytes.
lvcreate-free logvol volgroup percentCreate an LVM logical volume called /dev/volgroup/logvol, using approximately percent % of the free space remaining in the volume group. Most usefully, when percent is 100 this will create the largest possible LV.
lvm-canonical-lv-name lvnameThis converts alternative naming schemes for LVs that you might find to the canonical name. For example, /dev/mapper/VG-LV is converted to /dev/VG/LV.
This command returns an error if the lvname parameter does not refer to a logical volume.
See also "is-lv", "canonical-device-name".
lvm-clear-filterThis undoes the effect of "lvm-set-filter". LVM will be able to see every block device.
This command also clears the LVM cache and performs a volume group scan.
lvm-remove-allThis command removes all LVM logical volumes, volume groups and physical volumes.
lvm-set-filter 'devices ...'This sets the LVM device filter so that LVM will only be able to "see" the block devices in the list devices, and will ignore all other attached block devices.
Where disk image(s) contain duplicate PVs or VGs, this command is useful to get LVM to ignore the duplicates, otherwise LVM can get confused. Note also there are two types of duplication possible: either cloned PVs/VGs which have identical UUIDs; or VGs that are not cloned but just happen to have the same name. In normal operation you cannot create this situation, but you can do it outside LVM, eg. by cloning disk images or by bit twiddling inside the LVM metadata.
This command also clears the LVM cache and performs a volume group scan.
You can filter whole block devices or individual partitions.
You cannot use this if any VG is currently in use (eg. contains a mounted filesystem), even if you are not filtering out that VG.
lvremove deviceRemove an LVM logical volume device, where device is the path to the LV, such as /dev/VG/LV.
You can also remove all LVs in a volume group by specifying the VG name, /dev/VG.
lvrename logvol newlogvolRename a logical volume logvol with the new name newlogvol.
lvresize device mbytesThis resizes (expands or shrinks) an existing LVM logical volume to mbytes. When reducing, data in the reduced part is lost.
lvresize-free lv percentThis expands an existing logical volume lv so that it fills pc% of the remaining free space in the volume group. Commonly you would call this with pc = 100 which expands the logical volume as much as possible, using all remaining free space in the volume group.
lvsList all the logical volumes detected. This is the equivalent of the lvs(8) command.
This returns a list of the logical volume device names (eg. /dev/VolGroup00/LogVol00).
See also "lvs-full", "list-filesystems".
lvs-fullList all the logical volumes detected. This is the equivalent of the lvs(8) command. The "full" version includes all fields.
lvuuid deviceThis command returns the UUID of the LVM LV device.
lxattrlist path 'names ...'This call allows you to get the extended attributes of multiple files, where all files are in the directory path. names is the list of files from this directory.
On return you get a flat list of xattr structs which must be interpreted sequentially. The first xattr struct always has a zero-length attrname. attrval in this struct is zero-length to indicate there was an error doing lgetxattr for this file, or is a C string which is a decimal number (the number of following attributes for this file, which could be "0"). Then after the first xattr struct are the zero or more attributes for the first named file. This repeats for the second and subsequent files.
This call is intended for programs that want to efficiently list a directory contents without making many round-trips. See also "lstatlist" for a similarly efficient call for getting standard stats.
max-disksReturn the maximum number of disks that may be added to a handle (eg. by "add-drive-opts" and similar calls).
This function was added in libguestfs 1.19.7. In previous versions of libguestfs the limit was 25.
See "MAXIMUM NUMBER OF DISKS" in guestfs(3) for additional information on this topic.
md-create name 'devices ...' [missingbitmap:N] [nrdevices:N] [spare:N] [chunk:N] [level:..]Create a Linux md (RAID) device named name on the devices in the list devices.
The optional parameters are:
missingbitmapA bitmap of missing devices. If a bit is set it means that a missing device is added to the array. The least significant bit corresponds to the first device in the array.
As examples:
If devices = ["/dev/sda"] and missingbitmap = 0x1 then the resulting array would be [<missing>, "/dev/sda"].
If devices = ["/dev/sda"] and missingbitmap = 0x2 then the resulting array would be ["/dev/sda", <missing>].
This defaults to 0 (no missing devices).
The length of devices + the number of bits set in missingbitmap must equal nrdevices + spare.
nrdevicesThe number of active RAID devices.
If not set, this defaults to the length of devices plus the number of bits set in missingbitmap.
spareThe number of spare devices.
If not set, this defaults to 0.
chunkThe chunk size in bytes.
levelThe RAID level, which can be one of: linear, raid0, 0, stripe, raid1, 1, mirror, raid4, 4, raid5, 5, raid6, 6, raid10, 10. Some of these are synonymous, and more levels may be added in future.
If not set, this defaults to raid1.
This command has one or more optional arguments. See "OPTIONAL ARGUMENTS".
md-detail mdThis command exposes the output of 'mdadm -DY <md>'. The following fields are usually present in the returned hash. Other fields may also be present.
levelThe raid level of the MD device.
devicesThe number of underlying devices in the MD device.
metadataThe metadata version used.
uuidThe UUID of the MD device.
nameThe name of the MD device.
md-stat mdThis call returns a list of the underlying devices which make up the single software RAID array device md.
To get a list of software RAID devices, call "list-md-devices".
Each structure returned corresponds to one device along with additional status information:
mdstat_deviceThe name of the underlying device.
mdstat_indexThe index of this device within the array.
mdstat_flagsFlags associated with this device. This is a string containing (in no specific order) zero or more of the following flags:
Wwrite-mostly
Fdevice is faulty
Sdevice is a RAID spare
Rreplacement
md-stop mdThis command deactivates the MD array named md. The device is stopped, but it is not destroyed or zeroed.
mkdir pathCreate a directory named path.
mkdir-mode path modeThis command creates a directory, setting the initial permissions of the directory to mode.
For common Linux filesystems, the actual mode which is set will be mode & ~umask & 01777. Non-native-Linux filesystems may interpret the mode in other ways.
mkdir-p pathCreate a directory named path, creating any parent directories as necessary. This is like the mkdir -p shell command.
mkdtemp tmplThis command creates a temporary directory. The tmpl parameter should be a full pathname for the temporary directory name with the final six characters being "XXXXXX".
For example: "/tmp/myprogXXXXXX" or "/Temp/myprogXXXXXX", the second one being suitable for Windows filesystems.
The name of the temporary directory that was created is returned.
The temporary directory is created with mode 0700 and is owned by root.
The caller is responsible for deleting the temporary directory and its contents after use.
See also: mkdtemp(3)
mke2fs device [blockscount:N] [blocksize:N] [fragsize:N] [blockspergroup:N] [numberofgroups:N] [bytesperinode:N] [inodesize:N] [journalsize:N] [numberofinodes:N] [stridesize:N] [stripewidth:N] [maxonlineresize:N] [reservedblockspercentage:N] [mmpupdateinterval:N] [journaldevice:..] [label:..] [lastmounteddir:..] [creatoros:..] [fstype:..] [usagetype:..] [uuid:..] [forcecreate:true|false] [writesbandgrouponly:true|false] [lazyitableinit:true|false] [lazyjournalinit:true|false] [testfs:true|false] [discard:true|false] [quotatype:true|false] [extent:true|false] [filetype:true|false] [flexbg:true|false] [hasjournal:true|false] [journaldev:true|false] [largefile:true|false] [quota:true|false] [resizeinode:true|false] [sparsesuper:true|false] [uninitbg:true|false]mke2fs is used to create an ext2, ext3, or ext4 filesystem on device.
The optional blockscount is the size of the filesystem in blocks. If omitted it defaults to the size of device. Note if the filesystem is too small to contain a journal, mke2fs will silently create an ext2 filesystem instead.
This command has one or more optional arguments. See "OPTIONAL ARGUMENTS".
mke2fs-J fstype blocksize device journalThis creates an ext2/3/4 filesystem on device with an external journal on journal. It is equivalent to the command:
mke2fs -t fstype -b blocksize -J device=<journal> <device>See also "mke2journal".
This function is deprecated. In new code, use the "mke2fs" 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.
mke2fs-JL fstype blocksize device labelThis creates an ext2/3/4 filesystem on device with an external journal on the journal labeled label.
See also "mke2journal-L".
This function is deprecated. In new code, use the "mke2fs" 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.
mke2fs-JU fstype blocksize device uuidThis creates an ext2/3/4 filesystem on device with an external journal on the journal with UUID uuid.
See also "mke2journal-U".
This function is deprecated. In new code, use the "mke2fs" 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.
mke2journal blocksize deviceThis creates an ext2 external journal on device. It is equivalent to the command:
mke2fs -O journal_dev -b blocksize deviceThis function is deprecated. In new code, use the "mke2fs" 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.
mke2journal-L blocksize label deviceThis creates an ext2 external journal on device with label label.
This function is deprecated. In new code, use the "mke2fs" 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.
mke2journal-U blocksize uuid deviceThis creates an ext2 external journal on device with UUID uuid.
This function is deprecated. In new code, use the "mke2fs" 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.
mkfifo mode pathThis call creates a FIFO (named pipe) called path with mode mode. It is just a convenient wrapper around "mknod".
The mode actually set is affected by the umask.
mkfs fstype device [blocksize:N] [features:..] [inode:N] [sectorsize:N]This function creates a filesystem on device. The filesystem type is fstype, for example ext3.
The optional arguments are:
blocksizeThe filesystem block size. Supported block sizes depend on the filesystem type, but typically they are 1024, 2048 or 4096 for Linux ext2/3 filesystems.
For VFAT and NTFS the blocksize parameter is treated as the requested cluster size.
For UFS block sizes, please see mkfs.ufs(8).
featuresThis passes the -O parameter to the external mkfs program.
For certain filesystem types, this allows extra filesystem features to be selected. See mke2fs(8) and mkfs.ufs(8) for more details.
You cannot use this optional parameter with the gfs or gfs2 filesystem type.
inodeThis passes the -I parameter to the external mke2fs(8) program which sets the inode size (only for ext2/3/4 filesystems at present).
sectorsizeThis passes the -S parameter to external mkfs.ufs(8) program, which sets sector size for ufs filesystem.
This command has one or more optional arguments. See "OPTIONAL ARGUMENTS".
mkfs-b fstype blocksize deviceThis call is similar to "mkfs", but it allows you to control the block size of the resulting filesystem. Supported block sizes depend on the filesystem type, but typically they are 1024, 2048 or 4096 only.
For VFAT and NTFS the blocksize parameter is treated as the requested cluster size.
This function is deprecated. In new code, use the "mkfs" 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.
mkfs-btrfs 'devices ...' [allocstart:N] [bytecount:N] [datatype:..] [leafsize:N] [label:..] [metadata:..] [nodesize:N] [sectorsize:N]Create a btrfs filesystem, allowing all configurables to be set. For more information on the optional arguments, see mkfs.btrfs(8).
Since btrfs filesystems can span multiple devices, this takes a non-empty list of devices.
To create general filesystems, use "mkfs".
This command has one or more optional arguments. See "OPTIONAL ARGUMENTS".
mklost-and-found mountpointMake the lost+found directory, normally in the root directory of an ext2/3/4 filesystem. mountpoint is the directory under which we try to create the lost+found directory.
mkmountpoint exemptpath"mkmountpoint" and "rmmountpoint" are specialized calls that can be used to create extra mountpoints before mounting the first filesystem.
These calls are only necessary in some very limited circumstances, mainly the case where you want to mount a mix of unrelated and/or read-only filesystems together.
For example, live CDs often contain a "Russian doll" nest of filesystems, an ISO outer layer, with a squashfs image inside, with an ext2/3 image inside that. You can unpack this as follows in guestfish:
add-ro Fedora-11-i686-Live.iso
run
mkmountpoint /cd
mkmountpoint /sqsh
mkmountpoint /ext3fs
mount /dev/sda /cd
mount-loop /cd/LiveOS/squashfs.img /sqsh
mount-loop /sqsh/LiveOS/ext3fs.img /ext3fsThe inner filesystem is now unpacked under the /ext3fs mountpoint.
"mkmountpoint" is not compatible with "umount-all". You may get unexpected errors if you try to mix these calls. It is safest to manually unmount filesystems and remove mountpoints after use.
"umount-all" unmounts filesystems by sorting the paths longest first, so for this to work for manual mountpoints, you must ensure that the innermost mountpoints have the longest pathnames, as in the example code above.
For more details see https://bugzilla.redhat.com/show_bug.cgi?id=599503
Autosync [see "set-autosync", this is set by default on handles] can cause "umount-all" to be called when the handle is closed which can also trigger these issues.
mknod mode devmajor devminor pathThis call creates block or character special devices, or named pipes (FIFOs).
The mode parameter should be the mode, using the standard constants. devmajor and devminor are the device major and minor numbers, only used when creating block and character special devices.
Note that, just like mknod(2), the mode must be bitwise OR'd with S_IFBLK, S_IFCHR, S_IFIFO or S_IFSOCK (otherwise this call just creates a regular file). These constants are available in the standard Linux header files, or you can use "mknod-b", "mknod-c" or "mkfifo" which are wrappers around this command which bitwise OR in the appropriate constant for you.
The mode actually set is affected by the umask.
mknod-b mode devmajor devminor pathThis call creates a block device node called path with mode mode and device major/minor devmajor and devminor. It is just a convenient wrapper around "mknod".
The mode actually set is affected by the umask.
mknod-c mode devmajor devminor pathThis call creates a char device node called path with mode mode and device major/minor devmajor and devminor. It is just a convenient wrapper around "mknod".
The mode actually set is affected by the umask.
mkswap device [label:..] [uuid:..]Create a Linux swap partition on device.
The option arguments label and uuid allow you to set the label and/or UUID of the new swap partition.
This command has one or more optional arguments. See "OPTIONAL ARGUMENTS".
mkswap-L label deviceCreate a swap partition on device with label label.
Note that you cannot attach a swap label to a block device (eg. /dev/sda), just to a partition. This appears to be a limitation of the kernel or swap tools.
This function is deprecated. In new code, use the "mkswap" 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.
mkswap-U uuid deviceCreate a swap partition on device with UUID uuid.
This function is deprecated. In new code, use the "mkswap" 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.
mkswap-file pathCreate a swap file.
This command just writes a swap file signature to an existing file. To create the file itself, use something like "fallocate".
mktemp tmpl [suffix:..]This command creates a temporary file. The tmpl parameter should be a full pathname for the temporary directory name with the final six characters being "XXXXXX".
For example: "/tmp/myprogXXXXXX" or "/Temp/myprogXXXXXX", the second one being suitable for Windows filesystems.
The name of the temporary file that was created is returned.
The temporary file is created with mode 0600 and is owned by root.
The caller is responsible for deleting the temporary file after use.
If the optional suffix parameter is given, then the suffix (eg. .txt) is appended to the temporary name.
See also: "mkdtemp".
This command has one or more optional arguments. See "OPTIONAL ARGUMENTS".
modprobe modulenameThis loads a kernel module in the appliance.
The kernel module must have been whitelisted when libguestfs was built (see appliance/kmod.whitelist.in in the source).
mount mountable mountpointMount a guest disk at a position in the filesystem. Block devices are named /dev/sda, /dev/sdb and so on, as they were added to the guest. If those block devices contain partitions, they will have the usual names (eg. /dev/sda1). Also LVM /dev/VG/LV-style names can be used, or 'mountable' strings returned by "list-filesystems" or "inspect-get-mountpoints".
The rules are the same as for mount(2): A filesystem must first be mounted on / before others can be mounted. Other filesystems can only be mounted on directories which already exist.
The mounted filesystem is writable, if we have sufficient permissions on the underlying device.
Before libguestfs 1.13.16, this call implicitly added the options sync and noatime. The sync option greatly slowed writes and caused many problems for users. If your program might need to work with older versions of libguestfs, use "mount-options" instead (using an empty string for the first parameter if you don't want any options).
mount-9p mounttag mountpoint [options:..]Mount the virtio-9p filesystem with the tag mounttag on the directory mountpoint.
If required, trans=virtio will be automatically added to the options. Any other options required can be passed in the optional options parameter.
This command has one or more optional arguments. See "OPTIONAL ARGUMENTS".
mount-local localmountpoint [readonly:true|false] [options:..] [cachetimeout:N] [debugcalls:true|false]This call exports the libguestfs-accessible filesystem to a local mountpoint (directory) called localmountpoint. Ordinary reads and writes to files and directories under localmountpoint are redirected through libguestfs.
If the optional readonly flag is set to true, then writes to the filesystem return error EROFS.
options is a comma-separated list of mount options. See guestmount(1) for some useful options.
cachetimeout sets the timeout (in seconds) for cached directory entries. The default is 60 seconds. See guestmount(1) for further information.
If debugcalls is set to true, then additional debugging information is generated for every FUSE call.
When "mount-local" returns, the filesystem is ready, but is not processing requests (access to it will block). You have to call "mount-local-run" to run the main loop.
See "MOUNT LOCAL" in guestfs(3) for full documentation.
This command has one or more optional arguments. See "OPTIONAL ARGUMENTS".
mount-local-runRun the main loop which translates kernel calls to libguestfs calls.
This should only be called after "mount-local" returns successfully. The call will not return until the filesystem is unmounted.
Note you must not make concurrent libguestfs calls on the same handle from another thread.
You may call this from a different thread than the one which called "mount-local", subject to the usual rules for threads and libguestfs (see "MULTIPLE HANDLES AND MULTIPLE THREADS" in guestfs(3)).
See "MOUNT LOCAL" in guestfs(3) for full documentation.
mount-loop file mountpointThis command lets you mount file (a filesystem image in a file) on a mount point. It is entirely equivalent to the command mount -o loop file mountpoint.
mount-options options mountable mountpointThis is the same as the "mount" command, but it allows you to set the mount options as for the mount(8) -o flag.
If the options parameter is an empty string, then no options are passed (all options default to whatever the filesystem uses).
mount-ro mountable mountpointThis is the same as the "mount" command, but it mounts the filesystem with the read-only (-o ro) flag.
mount-vfs options vfstype mountable mountpointThis is the same as the "mount" command, but it allows you to set both the mount options and the vfstype as for the mount(8) -o and -t flags.
mountpointsThis call is similar to "mounts". That call returns a list of devices. This one returns a hash table (map) of device name to directory where the device is mounted.
mountsThis returns the list of currently mounted filesystems. It returns the list of devices (eg. /dev/sda1, /dev/VG/LV).
Some internal mounts are not shown.
See also: "mountpoints"
mv src destThis moves a file from src to dest where dest is either a destination filename or destination directory.
See also: "rename".
nr-devicesThis returns the number of whole block devices that were added. This is the same as the number of devices that would be returned if you called "list-devices".
To find out the maximum number of devices that could be added, call "max-disks".
ntfs-3g-probe true|false deviceThis command runs the ntfs-3g.probe(8) command which probes an NTFS device for mountability. (Not all NTFS volumes can be mounted read-write, and some cannot be mounted at all).
rw is a boolean flag. Set it to true if you want to test if the volume can be mounted read-write. Set it to false if you want to test if the volume can be mounted read-only.
The return value is an integer which 0 if the operation would succeed, or some non-zero value documented in the ntfs-3g.probe(8) manual page.
ntfsclone-in (backupfile|-) deviceRestore the backupfile (from a previous call to "ntfsclone-out") to device, overwriting any existing contents of this device.
Use - instead of a filename to read/write from stdin/stdout.
ntfsclone-out device (backupfile|-) [metadataonly:true|false] [rescue:true|false] [ignorefscheck:true|false] [preservetimestamps:true|false] [force:true|false]Stream the NTFS filesystem device to the local file backupfile. The format used for the backup file is a special format used by the ntfsclone(8) tool.
If the optional metadataonly flag is true, then only the metadata is saved, losing all the user data (this is useful for diagnosing some filesystem problems).
The optional rescue, ignorefscheck, preservetimestamps and force flags have precise meanings detailed in the ntfsclone(8) man page.
Use "ntfsclone-in" to restore the file back to a libguestfs device.
Use - instead of a filename to read/write from stdin/stdout.
This command has one or more optional arguments. See "OPTIONAL ARGUMENTS".
ntfsfix device [clearbadsectors:true|false]This command repairs some fundamental NTFS inconsistencies, resets the NTFS journal file, and schedules an NTFS consistency check for the first boot into Windows.
This is not an equivalent of Windows chkdsk. It does not scan the filesystem for inconsistencies.
The optional clearbadsectors flag clears the list of bad sectors. This is useful after cloning a disk with bad sectors to a new disk.
This command has one or more optional arguments. See "OPTIONAL ARGUMENTS".
ntfsresize device [size:N] [force:true|false]This command resizes an NTFS filesystem, expanding or shrinking it to the size of the underlying device.
The optional parameters are:
sizeThe new size (in bytes) of the filesystem. If omitted, the filesystem is resized to fit the container (eg. partition).
forceIf this option is true, then force the resize of the filesystem even if the filesystem is marked as requiring a consistency check.
After the resize operation, the filesystem is always marked as requiring a consistency check (for safety). You have to boot into Windows to perform this check and clear this condition. If you don't set the force option then it is not possible to call "ntfsresize" multiple times on a single filesystem without booting into Windows between each resize.
See also ntfsresize(8).
This command has one or more optional arguments. See "OPTIONAL ARGUMENTS".
ntfsresize-size device sizeThis command is the same as "ntfsresize" except that it allows you to specify the new size (in bytes) explicitly.
This function is deprecated. In new code, use the "ntfsresize" 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.
parse-environmentParse the program's environment and set flags in the handle accordingly. For example if LIBGUESTFS_DEBUG=1 then the 'verbose' flag is set in the handle.
Most programs do not need to call this. It is done implicitly when you call "create".
See "ENVIRONMENT VARIABLES" in guestfs(3) for a list of environment variables that can affect libguestfs handles. See also "guestfs_create_flags" in guestfs(3), and "parse-environment-list".
parse-environment-list 'environment ...'Parse the list of strings in the argument environment and set flags in the handle accordingly. For example if LIBGUESTFS_DEBUG=1 is a string in the list, then the 'verbose' flag is set in the handle.
This is the same as "parse-environment" except that it parses an explicit list of strings instead of the program's environment.
part-add device prlogex startsect endsectThis command adds a partition to device. If there is no partition table on the device, call "part-init" first.
The prlogex parameter is the type of partition. Normally you should pass p or primary here, but MBR partition tables also support l (or logical) and e (or extended) partition types.
startsect and endsect are the start and end of the partition in sectors. endsect may be negative, which means it counts backwards from the end of the disk (-1 is the last sector).
Creating a partition which covers the whole disk is not so easy. Use "part-disk" to do that.
part-del device partnumThis command deletes the partition numbered partnum on device.
Note that in the case of MBR partitioning, deleting an extended partition also deletes any logical partitions it contains.
part-disk device parttypeThis command is simply a combination of "part-init" followed by "part-add" to create a single primary partition covering the whole disk.
parttype is the partition table type, usually mbr or gpt, but other possible values are described in "part-init".
part-get-bootable device partnumThis command returns true if the partition partnum on device has the bootable flag set.
See also "part-set-bootable".
part-get-gpt-type device partnumReturn the type GUID of numbered GPT partition partnum. For MBR partitions, return an appropriate GUID corresponding to the MBR type. Behaviour is undefined for other partition types.
part-get-mbr-id device partnumReturns the MBR type byte (also known as the ID byte) from the numbered partition partnum.
Note that only MBR (old DOS-style) partitions have type bytes. You will get undefined results for other partition table types (see "part-get-parttype").
part-get-parttype deviceThis command examines the partition table on device and returns the partition table type (format) being used.
Common return values include: msdos (a DOS/Windows style MBR partition table), gpt (a GPT/EFI-style partition table). Other values are possible, although unusual. See "part-init" for a full list.
part-init device parttypeThis creates an empty partition table on device of one of the partition types listed below. Usually parttype should be either msdos or gpt (for large disks).
Initially there are no partitions. Following this, you should call "part-add" for each partition required.
Possible values for parttype are:
Intel EFI / GPT partition table.
This is recommended for >= 2 TB partitions that will be accessed from Linux and Intel-based Mac OS X. It also has limited backwards compatibility with the mbr format.
The standard PC "Master Boot Record" (MBR) format used by MS-DOS and Windows. This partition type will only work for device sizes up to 2 TB. For large disks we recommend using gpt.
Other partition table types that may work but are not supported include:
AIX disk labels.
Amiga "Rigid Disk Block" format.
BSD disk labels.
DASD, used on IBM mainframes.
MIPS/SGI volumes.
Old Mac partition format. Modern Macs use gpt.
NEC PC-98 format, common in Japan apparently.
Sun disk labels.
part-list deviceThis command parses the partition table on device and returns the list of partitions found.
The fields in the returned structure are:
Partition number, counting from 1.
Start of the partition in bytes. To get sectors you have to divide by the device's sector size, see "blockdev-getss".
End of the partition in bytes.
Size of the partition in bytes.
part-set-bootable device partnum true|falseThis sets the bootable flag on partition numbered partnum on device device. Note that partitions are numbered from 1.
The bootable flag is used by some operating systems (notably Windows) to determine which partition to boot from. It is by no means universally recognized.
part-set-gpt-type device partnum guidSet the type GUID of numbered GPT partition partnum to guid. Return an error if the partition table of device isn't GPT, or if guid is not a valid GUID.
See http://en.wikipedia.org/wiki/GUID_Partition_Table#Partition_type_GUIDs for a useful list of type GUIDs.
part-set-mbr-id device partnum idbyteSets the MBR type byte (also known as the ID byte) of the numbered partition partnum to idbyte. Note that the type bytes quoted in most documentation are in fact hexadecimal numbers, but usually documented without any leading "0x" which might be confusing.
Note that only MBR (old DOS-style) partitions have type bytes. You will get undefined results for other partition table types (see "part-get-parttype").
part-set-name device partnum nameThis sets the partition name on partition numbered partnum on device device. Note that partitions are numbered from 1.
The partition name can only be set on certain types of partition table. This works on gpt but not on mbr partitions.
part-to-dev partitionThis function takes a partition name (eg. "/dev/sdb1") and removes the partition number, returning the device name (eg. "/dev/sdb").
The named partition must exist, for example as a string returned from "list-partitions".
See also "part-to-partnum", "device-index".
part-to-partnum partitionThis function takes a partition name (eg. "/dev/sdb1") and returns the partition number (eg. 1).
The named partition must exist, for example as a string returned from "list-partitions".
See also "part-to-dev".
ping-daemonThis is a test probe into the guestfs daemon running inside the hypervisor. Calling this function checks that the daemon responds to the ping message, without affecting the daemon or attached block device(s) in any other way.
pread path count offsetThis command lets you read part of a file. It reads count bytes of the file, starting at offset, from file path.
This may read fewer bytes than requested. For further details see the pread(2) system call.
See also "pwrite", "pread-device".
Because of the message protocol, there is a transfer limit of somewhere between 2MB and 4MB. See "PROTOCOL LIMITS" in guestfs(3).
pread-device device count offsetThis command lets you read part of a block device. It reads count bytes of device, starting at offset.
This may read fewer bytes than requested. For further details see the pread(2) system call.
See also "pread".
Because of the message protocol, there is a transfer limit of somewhere between 2MB and 4MB. See "PROTOCOL LIMITS" in guestfs(3).
pvchange-uuid deviceGenerate a new random UUID for the physical volume device.
pvchange-uuid-allGenerate new random UUIDs for all physical volumes.
pvcreate deviceThis creates an LVM physical volume on the named device, where device should usually be a partition name such as /dev/sda1.
pvremove deviceThis wipes a physical volume device so that LVM will no longer recognise it.
The implementation uses the pvremove command which refuses to wipe physical volumes that contain any volume groups, so you have to remove those first.
pvresize deviceThis resizes (expands or shrinks) an existing LVM physical volume to match the new size of the underlying device.
pvresize-size device sizeThis command is the same as "pvresize" except that it allows you to specify the new size (in bytes) explicitly.
pvsList all the physical volumes detected. This is the equivalent of the pvs(8) command.
This returns a list of just the device names that contain PVs (eg. /dev/sda2).
See also "pvs-full".
pvs-fullList all the physical volumes detected. This is the equivalent of the pvs(8) command. The "full" version includes all fields.
pvuuid deviceThis command returns the UUID of the LVM PV device.
pwrite path content offsetThis command writes to part of a file. It writes the data buffer content to the file path starting at offset offset.
This command implements the pwrite(2) system call, and like that system call it may not write the full data requested. The return value is the number of bytes that were actually written to the file. This could even be 0, although short writes are unlikely for regular files in ordinary circumstances.
See also "pread", "pwrite-device".
Because of the message protocol, there is a transfer limit of somewhere between 2MB and 4MB. See "PROTOCOL LIMITS" in guestfs(3).
pwrite-device device content offsetThis command writes to part of a device. It writes the data buffer content to device starting at offset offset.
This command implements the pwrite(2) system call, and like that system call it may not write the full data requested (although short writes to disk devices and partitions are probably impossible with standard Linux kernels).
See also "pwrite".
Because of the message protocol, there is a transfer limit of somewhere between 2MB and 4MB. See "PROTOCOL LIMITS" in guestfs(3).
read-file pathThis calls returns the contents of the file path as a buffer.
Unlike "cat", this function can correctly handle files that contain embedded ASCII NUL characters.
read-lines pathReturn the contents of the file named path.
The file contents are returned as a list of lines. Trailing LF and CRLF character sequences are not returned.
Note that this function cannot correctly handle binary files (specifically, files containing \0 character which is treated as end of string). For those you need to use the "read-file" function and split the buffer into lines yourself.
readdir dirThis returns the list of directory entries in directory dir.
All entries in the directory are returned, including . and ... The entries are not sorted, but returned in the same order as the underlying filesystem.
Also this call returns basic file type information about each file. The ftyp field will contain one of the following characters:
Block special
Char special
Directory
FIFO (named pipe)
Symbolic link
Regular file
Socket
Unknown file type
The readdir(3) call returned a d_type field with an unexpected value
This function is primarily intended for use by programs. To get a simple list of names, use "ls". To get a printable directory for human consumption, use "ll".
Because of the message protocol, there is a transfer limit of somewhere between 2MB and 4MB. See "PROTOCOL LIMITS" in guestfs(3).
readlink pathThis command reads the target of a symbolic link.
readlinklist path 'names ...'This call allows you to do a readlink operation on multiple files, where all files are in the directory path. names is the list of files from this directory.
On return you get a list of strings, with a one-to-one correspondence to the names list. Each string is the value of the symbolic link.
If the readlink(2) operation fails on any name, then the corresponding result string is the empty string "". However the whole operation is completed even if there were readlink(2) errors, and so you can call this function with names where you don't know if they are symbolic links already (albeit slightly less efficient).
This call is intended for programs that want to efficiently list a directory contents without making many round-trips.
realpath pathReturn the canonicalized absolute pathname of path. The returned path has no ., .. or symbolic link path elements.
remount mountpoint [rw:true|false]This call allows you to change the rw (readonly/read-write) flag on an already mounted filesystem at mountpoint, converting a readonly filesystem to be read-write, or vice-versa.
Note that at the moment you must supply the "optional" rw parameter. In future we may allow other flags to be adjusted.
This command has one or more optional arguments. See "OPTIONAL ARGUMENTS".
remove-drive labelThis function is conceptually the opposite of "add-drive-opts". It removes the drive that was previously added with label label.
Note that in order to remove drives, you have to add them with labels (see the optional label argument to "add-drive-opts"). If you didn't use a label, then they cannot be removed.
You can call this function before or after launching the handle. If called after launch, if the backend supports it, we try to hot unplug the drive: see "HOTPLUGGING" in guestfs(3). The disk must not be in use (eg. mounted) when you do this. We try to detect if the disk is in use and stop you from doing this.
removexattr xattr pathThis call removes the extended attribute named xattr of the file path.
See also: "lremovexattr", attr(5).
rename oldpath newpathRename a file to a new place on the same filesystem. This is the same as the Linux rename(2) system call. In most cases you are better to use "mv" instead.
resize2fs deviceThis resizes an ext2, ext3 or ext4 filesystem to match the size of the underlying device.
See also "RESIZE2FS ERRORS" in guestfs(3).
resize2fs-M deviceThis command is the same as "resize2fs", but the filesystem is resized to its minimum size. This works like the -M option to the resize2fs command.
To get the resulting size of the filesystem you should call "tune2fs-l" and read the Block size and Block count values. These two numbers, multiplied together, give the resulting size of the minimal filesystem in bytes.
See also "RESIZE2FS ERRORS" in guestfs(3).
resize2fs-size device sizeThis command is the same as "resize2fs" except that it allows you to specify the new size (in bytes) explicitly.
See also "RESIZE2FS ERRORS" in guestfs(3).
rm pathRemove the single file path.
rm-f pathRemove the file path.
If the file doesn't exist, that error is ignored. (Other errors, eg. I/O errors or bad paths, are not ignored)
This call cannot remove directories. Use "rmdir" to remove an empty directory, or "rm-rf" to remove directories recursively.
rm-rf pathRemove the file or directory path, recursively removing the contents if its a directory. This is like the rm -rf shell command.
rmdir pathRemove the single directory path.
rmmountpoint exemptpathThis calls removes a mountpoint that was previously created with "mkmountpoint". See "mkmountpoint" for full details.
rsync src dest [archive:true|false] [deletedest:true|false]This call may be used to copy or synchronize two directories under the same libguestfs handle. This uses the rsync(1) program which uses a fast algorithm that avoids copying files unnecessarily.
src and dest are the source and destination directories. Files are copied from src to dest.
The optional arguments are:
archiveTurns on archive mode. This is the same as passing the --archive flag to rsync.
deletedestDelete files at the destination that do not exist at the source.
This command has one or more optional arguments. See "OPTIONAL ARGUMENTS".
rsync-in remote dest [archive:true|false] [deletedest:true|false]This call may be used to copy or synchronize the filesystem on the host or on a remote computer with the filesystem within libguestfs. This uses the rsync(1) program which uses a fast algorithm that avoids copying files unnecessarily.
This call only works if the network is enabled. See "set-network" or the --network option to various tools like guestfish(1).
Files are copied from the remote server and directory specified by remote to the destination directory dest.
The format of the remote server string is defined by rsync(1). Note that there is no way to supply a password or passphrase so the target must be set up not to require one.
The optional arguments are the same as those of "rsync".
This command has one or more optional arguments. See "OPTIONAL ARGUMENTS".
rsync-out src remote [archive:true|false] [deletedest:true|false]This call may be used to copy or synchronize the filesystem within libguestfs with a filesystem on the host or on a remote computer. This uses the rsync(1) program which uses a fast algorithm that avoids copying files unnecessarily.
This call only works if the network is enabled. See "set-network" or the --network option to various tools like guestfish(1).
Files are copied from the source directory src to the remote server and directory specified by remote.
The format of the remote server string is defined by rsync(1). Note that there is no way to supply a password or passphrase so the target must be set up not to require one.
The optional arguments are the same as those of "rsync".
Globbing does not happen on the src parameter. In programs which use the API directly you have to expand wildcards yourself (see "glob-expand"). In guestfish you can use the glob command (see "glob" in guestfish(1)), for example:
><fs> glob rsync-out /* rsync://remote/This command has one or more optional arguments. See "OPTIONAL ARGUMENTS".
scrub-device deviceThis command writes patterns over device to make data retrieval more difficult.
It is an interface to the scrub(1) program. See that manual page for more details.
scrub-file fileThis command writes patterns over a file to make data retrieval more difficult.
The file is removed after scrubbing.
It is an interface to the scrub(1) program. See that manual page for more details.
scrub-freespace dirThis command creates the directory dir and then fills it with files until the filesystem is full, and scrubs the files as for "scrub-file", and deletes them. The intention is to scrub any free space on the partition containing dir.
It is an interface to the scrub(1) program. See that manual page for more details.
set-append appendThis function is used to add additional options to the guest kernel command line.
The default is NULL unless overridden by setting LIBGUESTFS_APPEND environment variable.
Setting append to NULL means no additional options are passed (libguestfs always adds a few of its own).
set-attach-method backendSet the method that libguestfs uses to connect to the backend guestfsd daemon.
This function is deprecated. In new code, use the "set-backend" 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.
set-autosync true|falseIf autosync is true, this enables autosync. Libguestfs will make a best effort attempt to make filesystems consistent and synchronized when the handle is closed (also if the program exits without closing handles).
This is enabled by default (since libguestfs 1.5.24, previously it was disabled by default).
set-backend backendSet the method that libguestfs uses to connect to the backend guestfsd daemon.
This handle property was previously called the "attach method".
set-cachedir cachedirSet the directory used by the handle to store the appliance cache, when using a supermin appliance. The appliance is cached and shared between all handles which have the same effective user ID.
The environment variables LIBGUESTFS_CACHEDIR and TMPDIR control the default value: If LIBGUESTFS_CACHEDIR is set, then that is the default. Else if TMPDIR is set, then that is the default. Else /var/tmp is the default.
set-direct true|falseIf the direct appliance mode flag is enabled, then stdin and stdout are passed directly through to the appliance once it is launched.
One consequence of this is that log messages aren't caught by the library and handled by "set-log-message-callback", but go straight to stdout.
You probably don't want to use this unless you know what you are doing.
The default is disabled.
set-e2attrs file attrs [clear:true|false]This sets or clears the file attributes attrs associated with the inode file.
attrs is a string of characters representing file attributes. See "get-e2attrs" for a list of possible attributes. Not all attributes can be changed.
If optional boolean clear is not present or false, then the attrs listed are set in the inode.
If clear is true, then the attrs listed are cleared in the inode.
In both cases, other attributes not present in the attrs string are left unchanged.
These attributes are only present when the file is located on an ext2/3/4 filesystem. Using this call on other filesystem types will result in an error.
This command has one or more optional arguments. See "OPTIONAL ARGUMENTS".
set-e2generation file generationThis sets the ext2 file generation of a file.
See "get-e2generation".
set-e2label device labelThis sets the ext2/3/4 filesystem label of the filesystem on device to label. Filesystem labels are limited to 16 characters.
You can use either "tune2fs-l" or "get-e2label" to return the existing label on a filesystem.
This function is deprecated. In new code, use the "set-label" 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.
set-e2uuid device uuidThis sets the ext2/3/4 filesystem UUID of the filesystem on device to uuid. The format of the UUID and alternatives such as clear, random and time are described in the tune2fs(8) manpage.
You can use "vfs-uuid" to return the existing UUID of a filesystem.
This function is deprecated. In new code, use the "set-uuid" 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.
set-hv hvSet the hypervisor binary that we will use. The hypervisor depends on the backend, but is usually the location of the qemu/KVM hypervisor. For the uml backend, it is the location of the linux or vmlinux binary.
The default is chosen when the library was compiled by the configure script.
You can also override this by setting the LIBGUESTFS_HV environment variable.
Note that you should call this function as early as possible after creating the handle. This is because some pre-launch operations depend on testing qemu features (by running qemu -help). If the qemu binary changes, we don't retest features, and so you might see inconsistent results. Using the environment variable LIBGUESTFS_HV is safest of all since that picks the qemu binary at the same time as the handle is created.
set-label mountable labelSet the filesystem label on mountable to label.
Only some filesystem types support labels, and libguestfs supports setting labels on only a subset of these.
Labels are limited to 16 bytes.
Labels are limited to 128 unicode characters.
The label is limited to 12 bytes. The filesystem must not be mounted when trying to set the label.
The label is limited to 256 bytes and some characters are not allowed. Setting the label on a btrfs subvolume will set the label on its parent filesystem. The filesystem must not be mounted when trying to set the label.
To read the label on a filesystem, call "vfs-label".
set-libvirt-requested-credential index credAfter requesting the index'th credential from the user, call this function to pass the answer back to libvirt.
See "LIBVIRT AUTHENTICATION" in guestfs(3) for documentation and example code.
set-libvirt-supported-credentials 'creds ...'Call this function before setting an event handler for GUESTFS_EVENT_LIBVIRT_AUTH, to supply the list of credential types that the program knows how to process.
The creds list must be a non-empty list of strings. Possible strings are:
usernameauthnamelanguagecnoncepassphraseechopromptnoechopromptrealmexternalSee libvirt documentation for the meaning of these credential types.
See "LIBVIRT AUTHENTICATION" in guestfs(3) for documentation and example code.
set-memsize memsizeThis sets the memory size in megabytes allocated to the hypervisor. This only has any effect if called before "launch".
You can also change this by setting the environment variable LIBGUESTFS_MEMSIZE before the handle is created.
For more information on the architecture of libguestfs, see guestfs(3).
set-network true|falseIf network is true, then the network is enabled in the libguestfs appliance. The default is false.
This affects whether commands are able to access the network (see "RUNNING COMMANDS" in guestfs(3)).
You must call this before calling "launch", otherwise it has no effect.
set-path searchpathSet the path that libguestfs searches for kernel and initrd.img.
The default is $libdir/guestfs unless overridden by setting LIBGUESTFS_PATH environment variable.
Setting path to NULL restores the default path.
set-pgroup true|falseIf pgroup is true, child processes are placed into their own process group.
The practical upshot of this is that signals like SIGINT (from users pressing ^C) won't be received by the child process.
The default for this flag is false, because usually you want ^C to kill the subprocess. Guestfish sets this flag to true when used interactively, so that ^C can cancel long-running commands gracefully (see "user-cancel").
set-program programSet the program name. This is an informative string which the main program may optionally set in the handle.
When the handle is created, the program name in the handle is set to the basename from argv[0]. If that was not possible, it is set to the empty string (but never NULL).
set-qemu hvSet the hypervisor binary (usually qemu) that we will use.
The default is chosen when the library was compiled by the configure script.
You can also override this by setting the LIBGUESTFS_HV environment variable.
Setting hv to NULL restores the default qemu binary.
Note that you should call this function as early as possible after creating the handle. This is because some pre-launch operations depend on testing qemu features (by running qemu -help). If the qemu binary changes, we don't retest features, and so you might see inconsistent results. Using the environment variable LIBGUESTFS_HV is safest of all since that picks the qemu binary at the same time as the handle is created.
This function is deprecated. In new code, use the "set-hv" 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.
set-recovery-proc true|falseIf this is called with the parameter false then "launch" does not create a recovery process. The purpose of the recovery process is to stop runaway hypervisor processes in the case where the main program aborts abruptly.
This only has any effect if called before "launch", and the default is true.
About the only time when you would want to disable this is if the main process will fork itself into the background ("daemonize" itself). In this case the recovery process thinks that the main program has disappeared and so kills the hypervisor, which is not very helpful.
set-selinux true|falseThis sets the selinux flag that is passed to the appliance at boot time. The default is selinux=0 (disabled).
Note that if SELinux is enabled, it is always in Permissive mode (enforcing=0).
For more information on the architecture of libguestfs, see guestfs(3).
set-smp smpChange the number of virtual CPUs assigned to the appliance. The default is 1. Increasing this may improve performance, though often it has no effect.
This function must be called before "launch".
set-tmpdir tmpdirSet the directory used by the handle to store temporary files.
The environment variables LIBGUESTFS_TMPDIR and TMPDIR control the default value: If LIBGUESTFS_TMPDIR is set, then that is the default. Else if TMPDIR is set, then that is the default. Else /tmp is the default.
set-trace true|falseIf the command trace flag is set to 1, then libguestfs calls, parameters and return values are traced.
If you want to trace C API calls into libguestfs (and other libraries) then possibly a better way is to use the external ltrace(1) command.
Command traces are disabled unless the environment variable LIBGUESTFS_TRACE is defined and set to 1.
Trace messages are normally sent to stderr, unless you register a callback to send them somewhere else (see "set-event-callback").
set-uuid device uuidSet the filesystem UIUD on device to label.
Only some filesystem types support setting UUIDs.
To read the UUID on a filesystem, call "vfs-uuid".
set-verbose true|falseIf verbose is true, this turns on verbose messages.
Verbose messages are disabled unless the environment variable LIBGUESTFS_DEBUG is defined and set to 1.
Verbose messages are normally sent to stderr, unless you register a callback to send them somewhere else (see "set-event-callback").
setcon contextThis sets the SELinux security context of the daemon to the string context.
See the documentation about SELINUX in guestfs(3).
setxattr xattr val vallen pathThis call sets the extended attribute named xattr of the file path to the value val (of length vallen). The value is arbitrary 8 bit data.
See also: "lsetxattr", attr(5).
sfdisk device cyls heads sectors 'lines ...'This is a direct interface to the sfdisk(8) program for creating partitions on block devices.
device should be a block device, for example /dev/sda.
cyls, heads and sectors are the number of cylinders, heads and sectors on the device, which are passed directly to sfdisk as the -C, -H and -S parameters. If you pass 0 for any of these, then the corresponding parameter is omitted. Usually for 'large' disks, you can just pass 0 for these, but for small (floppy-sized) disks, sfdisk (or rather, the kernel) cannot work out the right geometry and you will need to tell it.
lines is a list of lines that we feed to sfdisk. For more information refer to the sfdisk(8) manpage.
To create a single partition occupying the whole disk, you would pass lines as a single element list, when the single element being the string , (comma).
See also: "sfdisk-l", "sfdisk-N", "part-init"
This function is deprecated. In new code, use the "part-add" 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.
sfdiskM device 'lines ...'This is a simplified interface to the "sfdisk" command, where partition sizes are specified in megabytes only (rounded to the nearest cylinder) and you don't need to specify the cyls, heads and sectors parameters which were rarely if ever used anyway.
See also: "sfdisk", the sfdisk(8) manpage and "part-disk"
This function is deprecated. In new code, use the "part-add" 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.
sfdisk-N device partnum cyls heads sectors lineThis runs sfdisk(8) option to modify just the single partition n (note: n counts from 1).
For other parameters, see "sfdisk". You should usually pass 0 for the cyls/heads/sectors parameters.
See also: "part-add"
This function is deprecated. In new code, use the "part-add" 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.
sfdisk-disk-geometry deviceThis displays the disk geometry of device read from the partition table. Especially in the case where the underlying block device has been resized, this can be different from the kernel's idea of the geometry (see "sfdisk-kernel-geometry").
The result is in human-readable format, and not designed to be parsed.
sfdisk-kernel-geometry deviceThis displays the kernel's idea of the geometry of device.
The result is in human-readable format, and not designed to be parsed.
sfdisk-l deviceThis displays the partition table on device, in the human-readable output of the sfdisk(8) command. It is not intended to be parsed.
See also: "part-list"
This function is deprecated. In new code, use the "part-list" 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.
sh commandThis call runs a command from the guest filesystem via the guest's /bin/sh.
This is like "command", but passes the command to:
/bin/sh -c "command"Depending on the guest's shell, this usually results in wildcards being expanded, shell expressions being interpolated and so on.
All the provisos about "command" apply to this call.
sh-lines commandThis is the same as "sh", but splits the result into a list of lines.
See also: "command-lines"
shutdownThis is the opposite of "launch". It performs an orderly shutdown of the backend process(es). If the autosync flag is set (which is the default) then the disk image is synchronized.
If the subprocess exits with an error then this function will return an error, which should not be ignored (it may indicate that the disk image could not be written out properly).
It is safe to call this multiple times. Extra calls are ignored.
This call does not close or free up the handle. You still need to call "close" afterwards.
"close" will call this if you don't do it explicitly, but note that any errors are ignored in that case.
sleep secsSleep for secs seconds.
stat pathReturns file information for the given path.
This is the same as the stat(2) system call.
statvfs pathReturns file system statistics for any mounted file system. path should be a file or directory in the mounted file system (typically it is the mount point itself, but it doesn't need to be).
This is the same as the statvfs(2) system call.
strings pathThis runs the strings(1) command on a file and returns the list of printable strings found.
Because of the message protocol, there is a transfer limit of somewhere between 2MB and 4MB. See "PROTOCOL LIMITS" in guestfs(3).
strings-e encoding pathThis is like the "strings" command, but allows you to specify the encoding of strings that are looked for in the source file path.
Allowed encodings are:
Single 7-bit-byte characters like ASCII and the ASCII-compatible parts of ISO-8859-X (this is what "strings" uses).
Single 8-bit-byte characters.
16-bit big endian strings such as those encoded in UTF-16BE or UCS-2BE.
16-bit little endian such as UTF-16LE and UCS-2LE. This is useful for examining binaries in Windows guests.
32-bit big endian such as UCS-4BE.
32-bit little endian such as UCS-4LE.
The returned strings are transcoded to UTF-8.
Because of the message protocol, there is a transfer limit of somewhere between 2MB and 4MB. See "PROTOCOL LIMITS" in guestfs(3).
swapoff-device deviceThis command disables the libguestfs appliance swap device or partition named device. See "swapon-device".
swapoff-file fileThis command disables the libguestfs appliance swap on file.
swapoff-label labelThis command disables the libguestfs appliance swap on labeled swap partition.
swapoff-uuid uuidThis command disables the libguestfs appliance swap partition with the given UUID.
swapon-device deviceThis command enables the libguestfs appliance to use the swap device or partition named device. The increased memory is made available for all commands, for example those run using "command" or "sh".
Note that you should not swap to existing guest swap partitions unless you know what you are doing. They may contain hibernation information, or other information that the guest doesn't want you to trash. You also risk leaking information about the host to the guest this way. Instead, attach a new host device to the guest and swap on that.
swapon-file fileThis command enables swap to a file. See "swapon-device" for other notes.
swapon-label labelThis command enables swap to a labeled swap partition. See "swapon-device" for other notes.
swapon-uuid uuidThis command enables swap to a swap partition with the given UUID. See "swapon-device" for other notes.
syncThis syncs the disk, so that any writes are flushed through to the underlying disk image.
You should always call this if you have modified a disk image, before closing the handle.
syslinux device [directory:..]Install the SYSLINUX bootloader on device.
The device parameter must be either a whole disk formatted as a FAT filesystem, or a partition formatted as a FAT filesystem. In the latter case, the partition should be marked as "active" ("part-set-bootable") and a Master Boot Record must be installed (eg. using "pwrite-device") on the first sector of the whole disk. The SYSLINUX package comes with some suitable Master Boot Records. See the syslinux(1) man page for further information.
The optional arguments are:
directoryInstall SYSLINUX in the named subdirectory, instead of in the root directory of the FAT filesystem.
Additional configuration can be supplied to SYSLINUX by placing a file called syslinux.cfg on the FAT filesystem, either in the root directory, or under directory if that optional argument is being used. For further information about the contents of this file, see syslinux(1).
See also "extlinux".
This command has one or more optional arguments. See "OPTIONAL ARGUMENTS".
tail pathThis command returns up to the last 10 lines of a file as a list of strings.
Because of the message protocol, there is a transfer limit of somewhere between 2MB and 4MB. See "PROTOCOL LIMITS" in guestfs(3).
tail-n nrlines pathIf the parameter nrlines is a positive number, this returns the last nrlines lines of the file path.
If the parameter nrlines is a negative number, this returns lines from the file path, starting with the -nrlinesth line.
If the parameter nrlines is zero, this returns an empty list.
Because of the message protocol, there is a transfer limit of somewhere between 2MB and 4MB. See "PROTOCOL LIMITS" in guestfs(3).
tar-in (tarfile|-) directory [compress:..]This command uploads and unpacks local file tarfile into directory.
The optional compress flag controls compression. If not given, then the input should be an uncompressed tar file. Otherwise one of the following strings may be given to select the compression type of the input file: compress, gzip, bzip2, xz, lzop. (Note that not all builds of libguestfs will support all of these compression types).
Use - instead of a filename to read/write from stdin/stdout.
This command has one or more optional arguments. See "OPTIONAL ARGUMENTS".
tar-out directory (tarfile|-) [compress:..] [numericowner:true|false] [excludes:..]This command packs the contents of directory and downloads it to local file tarfile.
The optional compress flag controls compression. If not given, then the output will be an uncompressed tar file. Otherwise one of the following strings may be given to select the compression type of the output file: compress, gzip, bzip2, xz, lzop. (Note that not all builds of libguestfs will support all of these compression types).
The other optional arguments are:
excludesA list of wildcards. Files are excluded if they match any of the wildcards.
numericownerIf set to true, the output tar file will contain UID/GID numbers instead of user/group names.
Use - instead of a filename to read/write from stdin/stdout.
This command has one or more optional arguments. See "OPTIONAL ARGUMENTS".
tgz-in (tarball|-) directoryThis command uploads and unpacks local file tarball (a gzip compressed tar file) into directory.
Use - instead of a filename to read/write from stdin/stdout.
This function is deprecated. In new code, use the "tar-in" 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.
tgz-out directory (tarball|-)This command packs the contents of directory and downloads it to local file tarball.
Use - instead of a filename to read/write from stdin/stdout.
This function is deprecated. In new code, use the "tar-out" 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.
touch pathTouch acts like the touch(1) command. It can be used to update the timestamps on a file, or, if the file does not exist, to create a new zero-length file.
This command only works on regular files, and will fail on other file types such as directories, symbolic links, block special etc.
truncate pathThis command truncates path to a zero-length file. The file must exist already.
truncate-size path sizeThis command truncates path to size size bytes. The file must exist already.
If the current file size is less than size then the file is extended to the required size with zero bytes. This creates a sparse file (ie. disk blocks are not allocated for the file until you write to it). To create a non-sparse file of zeroes, use "fallocate64" instead.
tune2fs device [force:true|false] [maxmountcount:N] [mountcount:N] [errorbehavior:..] [group:N] [intervalbetweenchecks:N] [reservedblockspercentage:N] [lastmounteddirectory:..] [reservedblockscount:N] [user:N]This call allows you to adjust various filesystem parameters of an ext2/ext3/ext4 filesystem called device.
The optional parameters are:
forceForce tune2fs to complete the operation even in the face of errors. This is the same as the tune2fs -f option.
maxmountcountSet the number of mounts after which the filesystem is checked by e2fsck(8). If this is 0 then the number of mounts is disregarded. This is the same as the tune2fs -c option.
mountcountSet the number of times the filesystem has been mounted. This is the same as the tune2fs -C option.
errorbehaviorChange the behavior of the kernel code when errors are detected. Possible values currently are: continue, remount-ro, panic. In practice these options don't really make any difference, particularly for write errors.
This is the same as the tune2fs -e option.
groupSet the group which can use reserved filesystem blocks. This is the same as the tune2fs -g option except that it can only be specified as a number.
intervalbetweenchecksAdjust the maximal time between two filesystem checks (in seconds). If the option is passed as 0 then time-dependent checking is disabled.
This is the same as the tune2fs -i option.
reservedblockspercentageSet the percentage of the filesystem which may only be allocated by privileged processes. This is the same as the tune2fs -m option.
lastmounteddirectorySet the last mounted directory. This is the same as the tune2fs -M option.
reservedblockscount Set the number of reserved filesystem blocks. This is the same as the tune2fs -r option.userSet the user who can use the reserved filesystem blocks. This is the same as the tune2fs -u option except that it can only be specified as a number.
To get the current values of filesystem parameters, see "tune2fs-l". For precise details of how tune2fs works, see the tune2fs(8) man page.
This command has one or more optional arguments. See "OPTIONAL ARGUMENTS".
tune2fs-l deviceThis returns the contents of the ext2, ext3 or ext4 filesystem superblock on device.
It is the same as running tune2fs -l device. See tune2fs(8) manpage for more details. The list of fields returned isn't clearly defined, and depends on both the version of tune2fs that libguestfs was built against, and the filesystem itself.
txz-in (tarball|-) directoryThis command uploads and unpacks local file tarball (an xz compressed tar file) into directory.
Use - instead of a filename to read/write from stdin/stdout.
This function is deprecated. In new code, use the "tar-in" 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.
txz-out directory (tarball|-)This command packs the contents of directory and downloads it to local file tarball (as an xz compressed tar archive).
Use - instead of a filename to read/write from stdin/stdout.
This function is deprecated. In new code, use the "tar-out" 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.
umask maskThis function sets the mask used for creating new files and device nodes to mask & 0777.
Typical umask values would be 022 which creates new files with permissions like "-rw-r--r--" or "-rwxr-xr-x", and 002 which creates new files with permissions like "-rw-rw-r--" or "-rwxrwxr-x".
The default umask is 022. This is important because it means that directories and device nodes will be created with 0644 or 0755 mode even if you specify 0777.
See also "get-umask", umask(2), "mknod", "mkdir".
This call returns the previous umask.
umount pathordevice [force:true|false] [lazyunmount:true|false]This unmounts the given filesystem. The filesystem may be specified either by its mountpoint (path) or the device which contains the filesystem.
This command has one or more optional arguments. See "OPTIONAL ARGUMENTS".
umount-allThis unmounts all mounted filesystems.
Some internal mounts are not unmounted by this call.
umount-local [retry:true|false]If libguestfs is exporting the filesystem on a local mountpoint, then this unmounts it.
See "MOUNT LOCAL" in guestfs(3) for full documentation.
This command has one or more optional arguments. See "OPTIONAL ARGUMENTS".
upload (filename|-) remotefilenameUpload local file filename to remotefilename on the filesystem.
filename can also be a named pipe.
See also "download".
Use - instead of a filename to read/write from stdin/stdout.
upload-offset (filename|-) remotefilename offsetUpload local file filename to remotefilename on the filesystem.
remotefilename is overwritten starting at the byte offset specified. The intention is to overwrite parts of existing files or devices, although if a non-existant file is specified then it is created with a "hole" before offset. The size of the data written is implicit in the size of the source filename.
Note that there is no limit on the amount of data that can be uploaded with this call, unlike with "pwrite", and this call always writes the full amount unless an error occurs.
Use - instead of a filename to read/write from stdin/stdout.
user-cancelThis function cancels the current upload or download operation.
Unlike most other libguestfs calls, this function is signal safe and thread safe. You can call it from a signal handler or from another thread, without needing to do any locking.
The transfer that was in progress (if there is one) will stop shortly afterwards, and will return an error. The errno (see "guestfs_last_errno") is set to EINTR, so you can test for this to find out if the operation was cancelled or failed because of another error.
No cleanup is performed: for example, if a file was being uploaded then after cancellation there may be a partially uploaded file. It is the caller's responsibility to clean up if necessary.
There are two common places that you might call "user-cancel":
In an interactive text-based program, you might call it from a SIGINT signal handler so that pressing ^C cancels the current operation. (You also need to call "guestfs_set_pgroup" so that child processes don't receive the ^C signal).
In a graphical program, when the main thread is displaying a progress bar with a cancel button, wire up the cancel button to call this function.
utimens path atsecs atnsecs mtsecs mtnsecsThis command sets the timestamps of a file with nanosecond precision.
atsecs, atnsecs are the last access time (atime) in secs and nanoseconds from the epoch.
mtsecs, mtnsecs are the last modification time (mtime) in secs and nanoseconds from the epoch.
If the *nsecs field contains the special value -1 then the corresponding timestamp is set to the current time. (The *secs field is ignored in this case).
If the *nsecs field contains the special value -2 then the corresponding timestamp is left unchanged. (The *secs field is ignored in this case).
utsnameThis returns the kernel version of the appliance, where this is available. This information is only useful for debugging. Nothing in the returned structure is defined by the API.
versionReturn the libguestfs version number that the program is linked against.
Note that because of dynamic linking this is not necessarily the version of libguestfs that you compiled against. You can compile the program, and then at runtime dynamically link against a completely different libguestfs.so library.
This call was added in version 1.0.58. In previous versions of libguestfs there was no way to get the version number. From C code you can use dynamic linker functions to find out if this symbol exists (if it doesn't, then it's an earlier version).
The call returns a structure with four elements. The first three (major, minor and release) are numbers and correspond to the usual version triplet. The fourth element (extra) is a string and is normally empty, but may be used for distro-specific information.
To construct the original version string: $major.$minor.$release$extra
See also: "LIBGUESTFS VERSION NUMBERS" in guestfs(3).
Note: Don't use this call to test for availability of features. In enterprise distributions we backport features from later versions into earlier versions, making this an unreliable way to test for features. Use "available" or "feature-available" instead.
vfs-label mountableThis returns the label of the filesystem on mountable.
If the filesystem is unlabeled, this returns the empty string.
To find a filesystem from the label, use "findfs-label".
vfs-type mountableThis command gets the filesystem type corresponding to the filesystem on mountable.
For most filesystems, the result is the name of the Linux VFS module which would be used to mount this filesystem if you mounted it without specifying the filesystem type. For example a string such as ext3 or ntfs.
vfs-uuid mountableThis returns the filesystem UUID of the filesystem on mountable.
If the filesystem does not have a UUID, this returns the empty string.
To find a filesystem from the UUID, use "findfs-uuid".
vg-activate true|false 'volgroups ...'This command activates or (if activate is false) deactivates all logical volumes in the listed volume groups volgroups.
This command is the same as running vgchange -a y|n volgroups...
Note that if volgroups is an empty list then all volume groups are activated or deactivated.
vg-activate-all true|falseThis command activates or (if activate is false) deactivates all logical volumes in all volume groups.
This command is the same as running vgchange -a y|n
vgchange-uuid vgGenerate a new random UUID for the volume group vg.
vgchange-uuid-allGenerate new random UUIDs for all volume groups.
vgcreate volgroup 'physvols ...'This creates an LVM volume group called volgroup from the non-empty list of physical volumes physvols.
vglvuuids vgnameGiven a VG called vgname, this returns the UUIDs of all the logical volumes created in this volume group.
You can use this along with "lvs" and "lvuuid" calls to associate logical volumes and volume groups.
See also "vgpvuuids".
vgmeta vgnamevgname is an LVM volume group. This command examines the volume group and returns its metadata.
Note that the metadata is an internal structure used by LVM, subject to change at any time, and is provided for information only.
vgpvuuids vgnameGiven a VG called vgname, this returns the UUIDs of all the physical volumes that this volume group resides on.
You can use this along with "pvs" and "pvuuid" calls to associate physical volumes and volume groups.
See also "vglvuuids".
vgremove vgnameRemove an LVM volume group vgname, (for example VG).
This also forcibly removes all logical volumes in the volume group (if any).
vgrename volgroup newvolgroupRename a volume group volgroup with the new name newvolgroup.
vgsList all the volumes groups detected. This is the equivalent of the vgs(8) command.
This returns a list of just the volume group names that were detected (eg. VolGroup00).
See also "vgs-full".
vgs-fullList all the volumes groups detected. This is the equivalent of the vgs(8) command. The "full" version includes all fields.
vgscanThis rescans all block devices and rebuilds the list of LVM physical volumes, volume groups and logical volumes.
vguuid vgnameThis command returns the UUID of the LVM VG named vgname.
wc-c pathThis command counts the characters in a file, using the wc -c external command.
wc-l pathThis command counts the lines in a file, using the wc -l external command.
wc-w pathThis command counts the words in a file, using the wc -w external command.
wipefs deviceThis command erases filesystem or RAID signatures from the specified device to make the filesystem invisible to libblkid.
This does not erase the filesystem itself nor any other data from the device.
Compare with "zero" which zeroes the first few blocks of a device.
write path contentThis call creates a file called path. The content of the file is the string content (which can contain any 8 bit data).
See also "write-append".
write-append path contentThis call appends content to the end of file path. If path does not exist, then a new file is created.
See also "write".
write-file path content sizeThis call creates a file called path. The contents of the file is the string content (which can contain any 8 bit data), with length size.
As a special case, if size is 0 then the length is calculated using strlen (so in this case the content cannot contain embedded ASCII NULs).
NB. Owing to a bug, writing content containing ASCII NUL characters does not work, even if the length is specified.
Because of the message protocol, there is a transfer limit of somewhere between 2MB and 4MB. See "PROTOCOL LIMITS" in guestfs(3).
This function is deprecated. In new code, use the "write" 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.
xfs-admin device [extunwritten:true|false] [imgfile:true|false] [v2log:true|false] [projid32bit:true|false] [lazycounter:true|false] [label:..] [uuid:..]Change the parameters of the XFS filesystem on device.
Devices that are mounted cannot be modified. Administrators must unmount filesystems before this call can modify parameters.
Some of the parameters of a mounted filesystem can be examined and modified using the "xfs-info" and "xfs-growfs" calls.
This command has one or more optional arguments. See "OPTIONAL ARGUMENTS".
xfs-growfs path [datasec:true|false] [logsec:true|false] [rtsec:true|false] [datasize:N] [logsize:N] [rtsize:N] [rtextsize:N] [maxpct:N]Grow the XFS filesystem mounted at path.
The returned struct contains geometry information. Missing fields are returned as -1 (for numeric fields) or empty string.
This command has one or more optional arguments. See "OPTIONAL ARGUMENTS".
xfs-info pathordevicepathordevice is a mounted XFS filesystem or a device containing an XFS filesystem. This command returns the geometry of the filesystem.
The returned struct contains geometry information. Missing fields are returned as -1 (for numeric fields) or empty string.
xfs-repair device [forcelogzero:true|false] [nomodify:true|false] [noprefetch:true|false] [forcegeometry:true|false] [maxmem:N] [ihashsize:N] [bhashsize:N] [agstride:N] [logdev:..] [rtdev:..]Repair corrupt or damaged XFS filesystem on device.
The filesystem is specified using the device argument which should be the device name of the disk partition or volume containing the filesystem. If given the name of a block device, xfs_repair will attempt to find the raw device associated with the specified block device and will use the raw device instead.
Regardless, the filesystem to be repaired must be unmounted, otherwise, the resulting filesystem may be inconsistent or corrupt.
The returned status indicates whether filesystem corruption was detected (returns 1) or was not detected (returns 0).
This command has one or more optional arguments. See "OPTIONAL ARGUMENTS".
zegrep regex pathThis calls the external zegrep program and returns the matching lines.
Because of the message protocol, there is a transfer limit of somewhere between 2MB and 4MB. See "PROTOCOL LIMITS" in guestfs(3).
This function is deprecated. In new code, use the "grep" 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.
zegrepi regex pathThis calls the external zegrep -i program and returns the matching lines.
Because of the message protocol, there is a transfer limit of somewhere between 2MB and 4MB. See "PROTOCOL LIMITS" in guestfs(3).
This function is deprecated. In new code, use the "grep" 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.
zero deviceThis command writes zeroes over the first few blocks of device.
How many blocks are zeroed isn't specified (but it's not enough to securely wipe the device). It should be sufficient to remove any partition tables, filesystem superblocks and so on.
If blocks are already zero, then this command avoids writing zeroes. This prevents the underlying device from becoming non-sparse or growing unnecessarily.
See also: "zero-device", "scrub-device", "is-zero-device"
zero-device deviceThis command writes zeroes over the entire device. Compare with "zero" which just zeroes the first few blocks of a device.
If blocks are already zero, then this command avoids writing zeroes. This prevents the underlying device from becoming non-sparse or growing unnecessarily.
zero-free-space directoryZero the free space in the filesystem mounted on directory. The filesystem must be mounted read-write.
The filesystem contents are not affected, but any free space in the filesystem is freed.
Free space is not "trimmed". You may want to call "fstrim" either as an alternative to this, or after calling this, depending on your requirements.
zerofree deviceThis runs the zerofree program on device. This program claims to zero unused inodes and disk blocks on an ext2/3 filesystem, thus making it possible to compress the filesystem more effectively.
You should not run this program if the filesystem is mounted.
It is possible that using this program can damage the filesystem or data on the filesystem.
zfgrep pattern pathThis calls the external zfgrep program and returns the matching lines.
Because of the message protocol, there is a transfer limit of somewhere between 2MB and 4MB. See "PROTOCOL LIMITS" in guestfs(3).
This function is deprecated. In new code, use the "grep" 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.
zfgrepi pattern pathThis calls the external zfgrep -i program and returns the matching lines.
Because of the message protocol, there is a transfer limit of somewhere between 2MB and 4MB. See "PROTOCOL LIMITS" in guestfs(3).
This function is deprecated. In new code, use the "grep" 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.
zfile meth pathThis command runs file after first decompressing path using method.
method must be one of gzip, compress or bzip2.
Since 1.0.63, use "file" instead which can now process compressed files.
This function is deprecated. In new code, use the "file" 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.
zgrep regex pathThis calls the external zgrep program and returns the matching lines.
Because of the message protocol, there is a transfer limit of somewhere between 2MB and 4MB. See "PROTOCOL LIMITS" in guestfs(3).
This function is deprecated. In new code, use the "grep" 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.
zgrepi regex pathThis calls the external zgrep -i program and returns the matching lines.
Because of the message protocol, there is a transfer limit of somewhere between 2MB and 4MB. See "PROTOCOL LIMITS" in guestfs(3).
This function is deprecated. In new code, use the "grep" 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.
guestfish returns 0 if the commands completed without error, or 1 if there was an error.
The edit command uses $EDITOR as the editor. If not set, it uses vi.
When using supermin ≥ 4.1.0, these have been renamed "SUPERMIN_KERNEL" and "SUPERMIN_MODULES".
The display command uses $GUESTFISH_DISPLAY_IMAGE to display images. If not set, it uses display(1).
Used with the --remote option to specify the remote guestfish process to control. See section "REMOTE CONTROL GUESTFISH OVER A SOCKET".
Set the command prompt. See "PROMPT".
The "hexedit" command uses $HEXEDITOR as the external hex editor. If not specified, the external hexedit(1) program is used.
If compiled with GNU readline support, various files in the home directory can be used. See "FILES".
Pass additional options to the guest kernel.
This is the old way to set LIBGUESTFS_BACKEND.
Choose the default way to create the appliance. See "guestfs_set_backend" in guestfs(3).
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".
Set LIBGUESTFS_DEBUG=1 to enable verbose messages. This has the same effect as using the -v option.
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.
Set the memory allocated to the qemu process, in megabytes. For example:
LIBGUESTFS_MEMSIZE=700Set the path that guestfish uses to search for kernel and initrd.img. See the discussion of paths in guestfs(3).
This is the old way to set LIBGUESTFS_HV.
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".
Set LIBGUESTFS_TRACE=1 to enable command traces.
The more command uses $PAGER as the pager. If not set, it uses more.
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.
These two 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-helper(8). This feature is only available in supermin / febootstrap ≥ 3.8.
This configuration file controls the default read-only or read-write mode (--ro or --rw).
If compiled with GNU readline support, then the command history is saved in this file.
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
...
$endifVariables that you can set in inputrc that change the behaviour of guestfish in useful ways include:
By default, guestfish will ignore case when tab-completing paths on the disk. Use:
set completion-ignore-case offto make guestfish case sensitive.
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.
guestfs(3), http://libguestfs.org/, virt-alignment-scan(1), virt-builder(1), virt-cat(1), virt-copy-in(1), virt-copy-out(1), virt-df(1), virt-edit(1), virt-filesystems(1), virt-inspector(1), virt-list-filesystems(1), virt-list-partitions(1), virt-ls(1), virt-make-fs(1), virt-rescue(1), virt-resize(1), virt-sparsify(1), virt-sysprep(1), virt-tar(1), virt-tar-in(1), virt-tar-out(1), virt-win-reg(1), libguestfs-tools.conf(5), display(1), hexedit(1), supermin-helper(8).
Richard W.M. Jones (rjones at redhat dot com)
Copyright (C) 2009-2014 Red Hat Inc.
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.
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.