VM

VM is just a set of scripts for myself that can be used to automate the startup and deployment of virtual machines. It is just a wrapper over the qemu command line.

System requirements

• qemu-system-* binary for your platform.
• qemu-img binary for creating initial HDD's.
• OVMF BIOS files for UEFI-booting in guest VM.
• pigz and unpigz binaries for snapshoting.
• remote-viewer binary (optional) for working with guest VM.
• bash on host and sh inside Rescue VM for executing jobs.

Features

• Creating, saving, restoring and destroying VM's.
• Many useful defaults for your VM's already set.
• Execution jobs-scripts inside ALT Rescue VM.
• Using color console and ability to tune colors.
• Optional using external media for store big data.
• Running on local computer with the graphics card.
• Running on remote server. SPICE can use in any case.
• Creating working files in place and in the $TMPDIR (default).

Preparing working space

• Copy scripts from bin to your binary directory and make it executable. You can modify this scripts as needed.
• Create sub-directory vm in your home directory and copy scripts sub-directory to.
• Prepare special rescue ISO-image for execution jobs inside ALT Rescue.
• Add the following lines to your ~/.ssh/config:

Host    kvm
        Hostname 127.0.0.1
        NoHostAuthenticationForLocalhost yes
        IdentityFile ~/.ssh/id_ed25519
        Port 5555
        User root

Preparing special rescue ISO-image

Here is sample action plan:

$ mkdir -p -m 755 ~/tmp/stage3 ~/vm/rescue-x86_64
$ cd ~/vm/rescue-x86_64/
$ wget http://nightly.altlinux.org/p10/permalink/alt-p10-rescue-latest-x86_64.iso
$ cp -Lf ~/vm/scripts/rescue-stage3 ~/tmp/stage3/autorun
$ alt2deploy -r ~/tmp/stage3 alt-p10-rescue-latest-x86_64.iso rescue-p10-x86_64.iso
$ rm -rf ~/tmp/stage3 alt-p10-rescue-latest-x86_64.iso
$ cat >guest.env <<EOF
RESCUE_CORES=4
RESCUE_MEMORY="2G"
RESCUE_ISO="rescue-p10-x86_64.iso"
RESCUE_FORWARDS="hostfwd=tcp::5555-:22"
EOF

Configuration and changing the defaults

• /etc/vm.conf - global settings for all users, they override the defaults.
• ~/.config/vm.conf - user personal setting, they also override the global settings for all users.
• vm.defaults - additional qemu arguments: this file can be placed in the /etc, ~/.config, $HOSTDIR or $WORKDIR directories (see vm-cmd-run.sh to get more details).

check-media script

The include script check-media must set the variable $MEDIA and, if necessary, make sure that the connected external media is available.

Example of ~/.config/vm.conf:

MEMORY=4G
MEDIA_MIRROR="/alt-mirror"
MEDIA_STORAGE="/images/iso"
FORWARDS="hostfwd=tcp::5555-:22"
SSHKEYS=( "$HOME/.ssh/id_ed25519.pub" )

Where:

• $MEDIA_MIRROR - Directory of repositories mirror relative to $MEDIA.
• $MEDIA_STORAGE - Directory of ISO-images relative to $MEDIA.
• $SSHKEYS (array) - List of public SSH-key paths that used in some scripts.

Creating snapshots

You can take as many snapshots as needed. The snapshot can be given an arbitrary name. You can add a short description to the snapshot. By default, snapshots are automatically named as S0, S1, S2, etc... The symlink LAST points to the last snapshot taken.

Restoring from a snapshot

The vm restore command allows you to restore the working state of the VM from the last snapshot taken. To restore from another snapshot, you should specify its name, ex.: vm restore 2023-05-10. If no working files are found when the VM starts up, a restore from the latest snapshot is automatically started.

Jobs

A job is an arbitrary set of files, one, two or three scripts, of which only script .job is required. Job scripts can be located in two places: in the VM directory and in the general location for storing them: $LIBEXEC/jobs. The rest of the files for a guest VM can be anywhere.

Usually script .pre copies everything needed to $WORKDIR/.in/, script .job is copied there too. After the job has completed and the virtual machine has shut down, the script .post will be run, if available.

Inside rescue VM you can use following mounted directories:

• /tmp/.mirror (optional, read only) - repositories mirror.
• /tmp/.in (read only) - for reading input data for the job.
• /tmp/.out (full access) - for saving job results and logs.

Inside <JOBNAME>.pre and <JOBNAME>.post scripts you can use all variables which declared in vm-main.sh. Most useful are:

• $HOSTDIR - primary directory with VM snapshots and settings.
• $WORKDIR - temporary directory with working files of the VM. It may be same as $HOSTDIR only when program running with the option --inplace, but by default working files created on the runfs or tmpfs (in the $TMPDIR).
• $LIBEXEC - directory with this utility scripts such as vm-main.sh.

Script <JOBNAME>.job running inside special rescue VM. It can use following variables:

• $MIRROR - repositories mirror (optional, read only).
• $INDIR - path to input data from the host system (read only).
• $OUTDIR - directory for saving results and logs (full access).
• $MNTOPTS - 9p-shares mount options for data exchange between host and guest systems.

add-ssh-keys job

This job copies all existing public SSH-keys, as specified by the $SSHKEYS variable, to the guest VM and adds each SSH-key for the superuser inside the installed guest OS. After completing this job you can login to the guest VM without a password prompt as superuser.

run-ssh job

This job also adds SSH-keys, but not to the installed system, but to a specialized rescue system, after which it raises the network in this rescue system, generates host keys and starts the SSH service. This allows further jobs to be executed via ssh immediately, no need to wait for the VM to start. To shut down a VM in a SSH-session, just remove the $WORKDIR/.out/RELEASE file and disconnect.

post-install job

Allows you to perform certain work after installing the OS:

• on the host system in your post-install.pre script and also copying the files to $INDIR;
• in ALT Rescue guest system with post-install.sh script;
• in the chroot of the installed OS from the same script;
• again on the host system in your post-install.post script.

The results are output to the terminal of the host system during the execution of all scripts inside ALT Rescue and are written to the $OUTDIR/job.log log. A typical usage is to have only one set of files in the virtual machine directory ($HOSTDIR) to run this job with the following default names:

• post-install.sh is the only required file for this job. Note that its name is determined by the $SETUP_SCRIPT variable, which can be changed through the passed set of parameters, or by the $SETUP_PROFILE variable, which can be exported through the environment when the command is run;
• post-install.env (optional) - a set of parameters passed to all executable scripts. There can be any arbitrary data, not just those described below;
• post-install.pre (optional) - executed on the host system before starting the main part of the job and can be used, for example, to check the possibility of its execution;
• post-install.post (optional) - executed on the host system after the completion of the main part of the task and can be used, for example, to process the results of the job;
• post-install.host-$SETUP_HOSTNAME.tgz (optional) - archive, usually unpacked by your script to the root of the installed OS. It is automatically copied to $INDIR if present and if $SETUP_HOSTNAME is set;
• post-install.user-$SETUP_USERNAME.tgz (optional) - archive, usually unpacked by your script to the user's home directory in the installed OS. It is automatically copied to $INDIR if present.

Note that there should not be a script named post-install.job in the $HOSTDIR directory, since it is already in $LIBEXEC/jobs and preparing your job, in particular, it mounts everything the necessary directories of the installed OS, so that in the future it would be easier to do chroot into it. There is no need to make files executable.

The post-install job can be run multiple times on the same virtual machine with a different set of files and parameters. To change the default values, export the $SETUP_PROFILE variable to the environment. The post-install job additionally defines the following parameters:

• $SETUP_PROFILE - Setup profile name. It is transmitted only through the environment. All of the files listed above begin with this name. Default is: "post-install".
• $SETUP_VERBOSE - Non-empty if the job in the guest OS with ALT Rescue should be run with debug information (set -x). Debugging is also enabled if the $INDIR/DEBUG file is present.
• $SETUP_ROOTDEV - Root partition of the installed OS, eg "/dev/sda3" or "LABEL=SYSTEM". If not specified, the mount-system command is used to automatically mount the directories of the installed OS.
• $SETUP_SCRIPT - The included script of the your job. By default, it's defined by the profile name with the addition of the ".sh" suffix.
• $SETUP_HOSTNAME - The name of the host with the installed OS, if it needs to be changed after installation, or to determine the name of an archive with files for unpack to the root of the installed OS.
• $SETUP_USERNAME - The name of the user in the installed OS to manipulate his profile. By default, the username is taken from the host OS.
• $SETUP_USER_UID - UID of the user in the installed OS for manipulating his profile. By default, the user's UID is taken from the host OS.
• $SETUP_FILES (array) - A list of files (relative to $HOSTDIR or with absolute paths) that need to be additionally copied to $INDIR before the job is executed.
• $DESTDIR - Directory with installed OS, for example, "/mnt/system1". Usually it is detected and mounted automatically, but if you need to perform these actions in your script in a different way, write your own non-empty value to this variable.

You can define any other parameters, for example:

• $SETUP_PASSWORD - Password of the regular user, if you need to change it after OS installation.
• $SETUP_ROOT_PWD - Administrator (root) password, if you need to change it after OS installation.
• $SETUP_TIMEZONE - Time zone, if you need to change it after OS installation.

You can use the following standard functions in your job script:

• fatal() - Displays an error message and crashes.
• run() - Performs the specified action and logs it.

See also example of the post-install usage in the builder sub-directory.

Usage examples

vm -c 8 -m 4G @create WS10 /iso/image.iso @job prepare @run @backup

VM files will be created in the current directory with the following parameters: GUESTNAME=WS10, CORES=8, MEMORY=4G, BOOTIMAGE=/iso/image.iso, all other by default; then job "prepare" will be executed inside special RESCUE VM; then VM will be bootted once from the specified BOOTIMAGE; and after the VM will be switched off, a snapshot "S0" will be created.

vm --no-uefi --tmeout=30 @restore S2 @job add-ssh-keys

VM will be restored from snapshot "S2"; then job "add-ssh-keys" will be executed inside special RESCUE VM: this job has a 30 seconds time limit for executing from job start; and VM will be started in Legacy/CSM mode, even if it VM was created with UEFI-boot.

Getting help

Just run: vm -h. See also: alt2deploy -h.

License

VM is licensed under the GNU General Public License (GPL), version 3, or (at your option) any later version.