1 Libguestfs is a library for accessing and modifying guest disk images.
2 Amongst the things this is good for: making batch configuration
3 changes to guests, getting disk used/free statistics (see also:
4 virt-df), migrating between virtualization systems (see also:
5 virt-p2v), performing partial backups, performing partial guest
6 clones, cloning guests and changing registry/UUID/hostname info, and
9 Libguestfs uses Linux kernel and qemu code, and can access any type of
10 guest filesystem that Linux and qemu can, including but not limited
11 to: ext2/3/4, btrfs, FAT and NTFS, LVM, many different disk partition
12 schemes, qcow, qcow2, vmdk.
14 Libguestfs provides ways to enumerate guest storage (eg. partitions,
15 LVs, what filesystem is in each LV, etc.). It can also run commands
16 in the context of the guest. Also you can access filesystems over
19 Libguestfs is a library that can be linked with C and C++ management
20 programs (or management programs written in OCaml, Perl, Python, Ruby,
21 Java, PHP, Haskell or C#). You can also use it from shell scripts or the
24 Libguestfs was written by Richard W.M. Jones (rjones@redhat.com) and
25 hacked on by lots of other people. For discussion, development,
26 patches, etc. please use the mailing list:
28 http://www.redhat.com/mailman/listinfo/libguestfs
32 ----------------------------------------------------------------------
34 http://libguestfs.org/
38 ----------------------------------------------------------------------
40 - recent QEMU >= 0.12 with virtio-serial support
48 - XDR, rpcgen (on Linux these are provided by glibc)
50 - pcre (Perl Compatible Regular Expressions C library) (optional)
52 - libmagic (the library that corresponds to the 'file' command) (optional)
58 - Augeas (http://augeas.net/) (optional)
62 - squashfs-tools (mksquashfs only)
64 - genisoimage / mkisofs
66 - hivex >= 1.2.1 (http://libguestfs.org/download)
68 - (Optional) FUSE to build the FUSE module
70 - perldoc (pod2man, pod2text) to generate the manual pages and
73 - (Optional) Readline to have nicer command-line editing in guestfish.
75 - (Optional) xmllint to validate virt-inspector RELAX NG schema
77 - (Optional) OCaml if you want to rebuild the generated files, and
78 also to build the OCaml bindings
80 - (Optional) local Fedora mirror
82 - (Optional) Perl if you want to build the perl bindings
84 - (Optional) Python if you want to build the python bindings
86 - (Optional) Ruby, rake if you want to build the ruby bindings
88 - (Optional) Java, JNI, jpackage-utils if you want to build the java
91 - (Optional) GHC if you want to build the Haskell bindings
93 - (Optional) Perl Sys::Virt module.
95 - (Optional) Perl Win::Hivex module.
97 - (Optional) Perl Pod::Usage module.
99 - (Optional) Perl Test::More module (from perl Test::Simple).
101 - (Optional, but highly recommended) perl-libintl for translating perl code.
103 - (Optional) po4a for translating manpages and POD files.
105 - (Optional) PHP, phpize if you want to build the PHP bindings
107 Running ./configure will check you have all the requirements installed
112 ----------------------------------------------------------------------
114 Then make the daemon, library and root filesystem:
116 ./configure [--with-mirror=URI]
119 Use the optional --with-mirror parameter to specify the URI of a local
120 Fedora mirror. See the discussion of the MIRROR parameter in the
121 febootstrap(8) manpage.
123 Finally run the tests:
127 If everything works, you can install the library and tools by running
128 this command as root:
134 ----------------------------------------------------------------------
136 We provide packages for Fedora >= 11 in Fedora. Use those, or build
137 from our source RPMs - it's far simpler that way.
139 You can compile libguestfs on Fedora 10 but you cannot use it with the
140 version of qemu in Fedora 10. You need to compile your own qemu, see
141 section 'qemu' below.
144 RHEL / EPEL / CentOS etc
145 ----------------------------------------------------------------------
147 We provide packages in EPEL which cover RHEL/CentOS >= 5. Use those
148 or build from our source RPMs.
152 ----------------------------------------------------------------------
154 libguestfs is now built as a package in Debian by Guido Gunther and
155 the other Debian libvirt maintainers. See:
157 http://wiki.debian.org/Teams/DebianLibvirtTeam#Packages
159 You can build for Debian in two different ways, either building a
160 Fedora-based appliance using febootstrap, yum, rpm, fakeroot,
161 fakechroot (all packaged in Debian). However the recommended way is
162 to build a Debian-based appliance using debootstrap and debirf.
164 Both ways are supported by the configure script.
168 ----------------------------------------------------------------------
170 By far the most common problem is with broken or incompatible
173 Different versions of qemu have problems booting the appliance for
174 different reasons. This varies between versions of qemu, and Linux
175 distributions which add their own patches.
177 If you find a problem, you could try using your own qemu built from
178 source (qemu is very easy to build from source), with a 'qemu
179 wrapper'. Qemu wrappers are described in the guestfs(3) manpage.
183 ----------------------------------------------------------------------
185 By default the configure script will look for qemu-kvm (KVM support).
186 You will need a reasonably recent processor for this to work. KVM is
187 much faster than using plain Qemu.
189 You may also need to enable KVM support for non-root users, by following
192 http://www.linux-kvm.org/page/FAQ#How_can_I_use_kvm_with_a_non-privileged_user.3F
194 On some systems, this will work too:
198 On some systems, the chmod will not survive a reboot, and you will
199 need to make edits to the udev configuration.
203 ----------------------------------------------------------------------
205 Previous versions of libguestfs required something called "vmchannel".
206 Vmchannel is a special device given to virtual machines which allows
207 them to communicate in some way with the host, often (but not always)
208 without using a traditional network device. In reality, there is no
209 one thing called "vmchannel". This idea has been reimplemented
210 several times under the name vmchannel, and other hypervisors have
211 their own incompatible implementation(s) too.
213 In libguestfs <= 1.0.71, we required a specific vmchannel which is
214 properly known as "guestfwd" and has been upstream in qemu since here:
216 http://lists.gnu.org/archive/html/qemu-devel/2009-02/msg01042.html
218 In libguestfs >= 1.0.71 we don't require any vmchannel implementation,
219 as long as qemu has been compiled with support for SLIRP (user mode
220 networking, or "-net user"), which is almost always the case.
222 In libguestfs >= 1.5.4 we switched again to using qemu's virtio-serial
223 and removed all the other vmchannels and the SLIRP channel.
227 ----------------------------------------------------------------------
229 If you configure with --enable-supermin then we will build a supermin
230 appliance (supermin = super-minimized). This is a very specialized
231 appliance which is built on-the-fly at runtime (specifically, when you
232 call guestfs_launch).
234 The normal appliance is a self-contained Linux operating system, based
235 on the Fedora/RHEL/CentOS Linux distro. So it contains a complete
236 copy of all the libraries and programs needed, like kernel, libc,
237 bash, coreutils etc etc.
239 The supermin appliance removes the kernel and all the executable
240 libraries and programs from the appliance. That just leaves a
241 skeleton of config files and some data files, which is obviously
242 massively smaller than the normal appliance. At runtime we rebuild
243 the appliance on-the-fly from the libraries and programs on the host
244 (eg. pulling in the real /lib/libc.so, the real /bin/bash etc.)
246 Although this process of rebuilding the appliance each time sounds
247 slow, it turns out to be faster than using the prebuilt appliance.
248 (Most of the saving comes from not compressing the appliance - it
249 transpires that decompressing the appliance is the slowest part of the
250 whole boot sequence). On my machine, a new appliance can be built in
251 under a fifth of a second, and the boot time is several seconds
254 The big advantage of the supermin appliance for distributions like
255 Fedora is that it gets security fixes automatically from the host, so
256 there is no need to rebuild the whole of libguestfs for a security
257 update in some underlying library.
259 There are several DISADVANTAGES:
261 It won't work at all except in very narrow, controlled cases like the
262 Fedora packaging case. We control the dependencies of the libguestfs
263 RPM tightly to ensure that the required binaries are actually present
266 Furthermore there are certain unlikely changes in the packages on the
267 host which could break a supermin appliance, eg. an updated library
268 which depends on an additional data file.
270 Also supermin appliances are subjected to changes in the host kernel
271 which might break compatibility with qemu -- these are, of course,
272 real bugs in any case.
274 Lastly, supermin appliances really can't be moved between branches of
275 distributions (eg. built on Fedora 12 and moved to Fedora 10) because
276 they are not self-contained and they rely on certain libraries being
277 around. You shouldn't do this anyway.
279 Use supermin appliances with caution.
282 Notes on cross-architecture support
283 ----------------------------------------------------------------------
285 At the moment we basically don't support cross-architecture or
286 32-on-64. This limits what is possible for some guests. Filesystem
287 operations and FUSE will work fine, but running commands in guests may
290 To enable this requires work for cross-architecture and 32-on-64
291 support in febootstrap, fakeroot and fakechroot.
293 The daemon/ directory contains its own configure script. This is so
294 that in future we will be able to cross-compile the daemon.
298 ----------------------------------------------------------------------
300 On my machines I can usually rebuild the appliance in around 3
301 minutes. If it takes much longer for you, use a local Fedora mirror
304 To use squid to cache yum downloads, read this first:
305 https://lists.dulug.duke.edu/pipermail/yum/2006-August/009041.html
306 (In brief, because yum chooses random mirrors each time, squid doesn't
307 work very well with default yum configuration. To get around this,
308 choose a Fedora mirror which is close to you, set this with
309 './configure --with-mirror=[...]', and then proxy the whole lot
310 through squid by setting http_proxy environment variable).
312 You will also need to substantially increase the squid configuration
314 http://fedoraproject.org/wiki/Using_Mock_to_test_package_builds#Using_Squid_to_Speed_Up_Mock_package_downloads
317 Porting to other Linux distros / non-Linux
318 ----------------------------------------------------------------------
320 libguestfs itself should be fairly portable to other Linux
321 distributions. Non-Linux ports are trickier, but we will accept
322 patches if they aren't too invasive.
324 The main porting issues are with the dependencies needed to build the
325 appliance. You will need to find or port the following packages
331 - rpm-python http://www.rpm.org/
332 - yum http://yum.baseurl.org/
333 - febootstrap http://people.redhat.com/~rjones/febootstrap/
336 Copyright and license information
337 ----------------------------------------------------------------------
339 Copyright (C) 2009-2010 Red Hat Inc.
341 The library is distributed under the LGPLv2+. The programs are
342 distributed under the GPLv2+. Please see the files COPYING and
343 COPYING.LIB for full license information.