virt-sysprep - Reset, unconfigure or customize a virtual machine so clones can be made
virt-sysprep [--options] -d domname virt-sysprep [--options] -a disk.img [-a disk.img ...]
Virt-sysprep can reset or unconfigure a virtual machine so that clones can be made from it. Steps in this process include removing SSH host keys, removing persistent network MAC configuration, and removing user accounts. Virt-sysprep can also customize a virtual machine, for instance by adding SSH keys, users or logos. Each step can be enabled or disabled as required.
Virt-sysprep modifies the guest or disk image in place. The guest must be shut down. If you want to preserve the existing contents of the guest, you must snapshot, copy or clone the disk first. See "COPYING AND CLONING" below.
You do not need to run virt-sysprep as root. In fact we'd generally recommend that you don't. The time you might want to run it as root is when you need root in order to access the disk image, but even in this case it would be better to change the permissions on the disk image to be writable as the non-root user running virt-sysprep.
"Sysprep" stands for "system preparation" tool. The name comes from the Microsoft program
sysprep.exe which is used to unconfigure Windows machines in preparation for cloning them. Having said that, virt-sysprep does not currently work on Microsoft Windows guests. We plan to support Windows sysprepping in a future version, and we already have code to do it.
Display brief help.
- -a file
- --add file
Add file which should be a disk image from a virtual machine.
The format of the disk image is auto-detected. To override this and force a particular format use the --format option.
- -c URI
- --connect URI
If using libvirt, connect to the given URI. If omitted, then we connect to the default libvirt hypervisor.
If you specify guest block devices directly (-a), then libvirt is not used at all.
- -d guest
- --domain guest
Add all the disks from the named libvirt guest. Domain UUIDs can be used instead of names.
Perform a read-only "dry run" on the guest. This runs the sysprep operation, but throws away any changes to the disk at the end.
- --enable operations
Choose which sysprep operations to perform. Give a comma-separated list of operations, for example:
would enable ONLY
If the --enable option is not given, then we default to trying most sysprep operations (see --list-operations to show which are enabled).
Regardless of the --enable option, sysprep operations are skipped for some guest types.
Use --list-operations to list operations supported by a particular version of virt-sysprep.
See "OPERATIONS" below for a list and an explanation of each operation.
- --format raw|qcow2|..
- --format auto
The default for the -a option is to auto-detect the format of the disk image. Using this forces the disk format for -a options which follow on the command line. Using --format auto switches back to auto-detection for subsequent -a options.
virt-sysprep --format raw -a disk.img
forces raw format (no auto-detection) for
virt-sysprep --format raw -a disk.img --format auto -a another.img
forces raw format (no auto-detection) for
disk.imgand reverts to auto-detection for
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).
List the operations supported by the virt-sysprep program.
These are listed one per line, with one or more single-space-separated fields, eg:
$ virt-sysprep --list-operations bash-history * Remove the bash history in the guest cron-spool * Remove user at-jobs and cron-jobs dhcp-client-state * Remove DHCP client leases dhcp-server-state * Remove DHCP server leases [etc]
The first field is the operation name, which can be supplied to --enable. The second field is a
*character if the operation is enabled by default or blank if not. Subsequent fields on the same line are the description of the operation.
Before libguestfs 1.17.33 only the first (operation name) field was shown and all operations were enabled by default.
Don't print log messages.
To enable detailed logging of individual file operations, use -x.
--selinux-relabel forces SELinux relabelling next time the guest boots. --no-selinux-relabel disables relabelling.
The default is to try to detect if SELinux relabelling is required. See "SELINUX RELABELLING" below for more details.
Enable verbose messages for debugging.
Display version number and exit.
Enable tracing of libguestfs API calls.
If the --enable option is not given, then most sysprep operations are enabled.
virt-sysprep --list-operations to list all operations for your virt-sysprep binary. The ones which are enabled by default are marked with a
* character. Regardless of the --enable option, sysprep operations are skipped for some guest types.
Operations can be individually enabled using the --enable option. Use a comma-separated list, for example:
virt-sysprep --enable=ssh-hostkeys,udev-persistent-net [etc..]
Future versions of virt-sysprep may add more operations. If you are using virt-sysprep and want predictable behaviour, specify only the operations that you want to have enabled.
* = enabled by default when no --enable option is given.
Virt-sysprep can be used as part of a process of cloning guests, or to prepare a template from which guests can be cloned. There are many different ways to achieve this using the virt tools, and this section is just an introduction.
A virtual machine (when switched off) consists of two parts:
The configuration or description of the guest. eg. The libvirt XML (see
virsh dumpxml), the running configuration of the guest, or another external format like OVF.
Some configuration items that might need to be changed:
- path to block device(s)
- network card MAC address
- block device(s)
One or more hard disk images, themselves containing files, directories, applications, kernels, configuration, etc.
Some things inside the block devices that might need to be changed:
- hostname and other net configuration
- SSH host keys
- Windows unique security ID (SID)
- Puppet registration
Starting with an original guest, you probably wish to copy the guest block device and its configuration to make a template. Then once you are happy with the template, you will want to make many clones from it.
virt-sysprep | v original guest --------> template ----------> \------> cloned \-----> guests \---->
dd dd original guest --------> template ----------> \------> cloned \-----> guests \---->
There are some smarter (and faster) ways too:
snapshot template ----------> \------> cloned \-----> guests \---->
You may want to run virt-sysprep twice, once to reset the guest (to make a template) and a second time to customize the guest for a specific user:
virt-sysprep virt-sysprep (reset) (add user, keys, logos) | | dd v dd v original guest ----> template ---------> copied ------> custom template guest
- Create a snapshot using qemu-img:
qemu-img create -f qcow2 -o backing_file=original snapshot.qcow
The advantage is that you don't need to copy the original (very fast) and only changes are stored (less storage required).
Note that writing to the backing file once you have created guests on top of it is not possible: you will corrupt the guests.
- Create a snapshot using
- Other ways to create snapshots include using filesystems-level tools (for filesystems such as btrfs).
Most Network Attached Storage (NAS) devices can also create cheap snapshots from files or LUNs.
- Get your NAS to duplicate the LUN. Most NAS devices can also duplicate LUNs very cheaply (they copy them on-demand in the background).
- Prepare your template using virt-sparsify(1). See below.
A separate tool, virt-clone(1), can be used to duplicate the block device and/or modify the external libvirt configuration of a guest. It will reset the name, UUID and MAC address of the guest in the libvirt XML.
virt-clone(1) does not use libguestfs and cannot look inside the disk image. This was the original motivation to write virt-sysprep.
virt-sparsify original guest --------> template
virt-sparsify(1) can be used to make the cloning template smaller, making it easier to compress and/or faster to copy.
Notice that since virt-sparsify also copies the image, you can use it to make the initial copy (instead of
virt-resize template ----------> \------> cloned \-----> guests \---->
If you want to give people cloned guests, but let them pick the size of the guest themselves (eg. depending on how much they are prepared to pay for disk space), then instead of copying the template, you can run virt-resize(1). Virt-resize performs a copy and resize, and thus is ideal for cloning guests from a template.
The two options --firstboot and --script both supply shell scripts that are run against the guest. However these two options are significantly different.
--firstboot script uploads the file
script into the guest and arranges that it will run, in the guest, when the guest is next booted. (The script will only run once, at the "first boot").
--script script runs the shell
script on the host, with its current directory inside the guest filesystem.
If you needed, for example, to
yum install new packages, then you must not use --script for this, since that would (a) run the
yum command on the host and (b) wouldn't have access to the same resources (repositories, keys, etc.) as the guest. Any command that needs to run on the guest must be run via --firstboot.
On the other hand if you need to make adjustments to the guest filesystem (eg. copying in files), then --script is ideal since (a) it has access to the host filesystem and (b) you will get immediate feedback on errors.
Either or both options can be used multiple times on the command line.
Although virt-sysprep removes some sensitive information from the guest, it does not pretend to remove all of it. You should examine the "OPERATIONS" above and the guest afterwards.
Sensitive files are simply removed. The data they contained may still exist on the disk, easily recovered with a hex editor or undelete tool. Use virt-sparsify(1) as one way to remove this content. See also the scrub(1) command to get rid of deleted content in directory entries and inodes.
(This section applies to Linux guests only)
random-seed operation writes a few bytes of randomness from the host into the guest's random seed file.
If this is just done once and the guest is cloned from the same template, then each guest will start with the same entropy, and things like SSH host keys and TCP sequence numbers may be predictable.
Therefore you should arrange to add more randomness after cloning from a template too, which can be done by just enabling the
cp template.img newguest.img virt-sysprep --enable random-seed -a newguest.img
(This section applies to Linux guests using SELinux only)
If any new files are created by virt-sysprep, then virt-sysprep touches
/.autorelabel so that these will be correctly labelled by SELinux the next time the guest is booted. This process interrupts boot and can take some time.
You can force relabelling for all guests by supplying the --selinux-relabel option.
You can disable relabelling entirely by supplying the --no-selinux-relabel option.
Libvirt guest names can contain arbitrary characters, some of which have meaning to the shell such as
# and space. You may need to quote or escape these characters on the command line. See the shell manual page sh(1) for details.
This program returns 0 on success, or 1 if there was an error.
Richard W.M. Jones http://people.redhat.com/~rjones/
Wanlong Gao, Fujitsu Ltd.
Copyright (C) 2011-2012 Red Hat Inc.
Copyright (C) 2012 Fujitsu Ltd.