...
Apptool is an OVF (open virtualization format) import tool. More info at http://wiki.cam.xci-test.com/index.php/Ovf <<< TODO that is a bad link - maybe make the page public?
XenVM cont.
There is some more documentation buried in the toolstack repository. I added it - for convenience purposes - here.
Xenvm is a single VM monitor. it's a single process that follow the life of a VM, and open a command socket where you can send command, just like xm to xend.
xenvm expose the uuid of the domain to outside world, so instead of having to refer to the domid of the domain, all outside command use the uuid. the uuid beeing stable, all command will use the same uuid after reboot/suspend/resume cycle.
Config file
you can put the following parameters in your config file:
(common options)
| key name | type | description |
|---|---|---|
| hvm | s | command line given to the kernel |
| startup | s | specify what to do with the domain at startup possibles value: started, paused, shutdown or restore |
| debug | b | logs all operations to /tmp/xenvm-debug-%uuid |
| uuid | s | specify the domain uuid (default to autogeneration) |
| on_crash | s | specify the action to be taken after notifying a |
| on_halt | s | crash/halt/reboot. possible values: |
| on_reboot | s | preserve, reboot, destroy |
| kernel | s | specify where to find the kernel to boot (can be empty for hvm. default to hvmloader) |
| memory | i | specify the memory given to the guest in megabytes |
| vcpus | i | number of vcpus available to the guest |
| disk | s | add a virtual disk. |
| format: physpath:phystype:virtpath:mode:devty[:k=v...] | ||
| - physpath: path to the disk image, raw device, .. | ||
| - phystype: phy | ||
| - virtpath: hd(a-d) | ||
| - mode: r | ||
| - devtype = cdrom | ||
| - extra k=v arguments (cipher, key-size, key-file) | ||
| nic | s | add a virtual nic. |
| format: key=value,key=value,... (can be empty) | ||
| supported key: bridge, mac, id | ||
| examples: | ||
| "nic = bridge=xen-br0,mac=ab:ef:fe:dc:ba:ab" | ||
| "nic = mac=ab:ef:fe:dc:ba:ab" | ||
| "nic = " | ||
| pci | s | add a pci device. |
| format: devid,bind,domain | ||
| serial | s | redirect serial to device or network tcp:ip:port |
| ex: "pty" or "tcp:1.2.3.4:1234" | ||
| display | s | details the type of display available for the guest |
| format: :key[=value],key[=value],... | ||
| possible values: | ||
| - none | ||
| - vnc (keys allowed: use-port-unused, keymap, port) | ||
| - sdl | ||
| - intel |
(the following are just useful for pv)
| key name | type | description |
|---|---|---|
| cmdline | s | command line given to the kernel |
| initrd | s | specify the initrd use, leave empty for none |
(the following are just useful for hvm)
| key name | type | description |
|---|---|---|
| pae | b | specify that the guest is using PAE |
| acpi | b | specify that the guest is using ACPI |
| apic | b | specify that the guest is using APIC |
| nx | b | specify that the guest is using NX |
| smbios-pt | b | specify that the guest is using smbios pass-through |
| smbios-oem-type-pt | i | tables number to passthrough |
| acpi-pt | b | specify that the guest is using ACPI pass-through |
| diskinfo-pt | b | specify the guest is using SCSI diskinfo pass-through |
| boot | s | specify the qemu boot string |
| extra-hvm | k=v | specify extra arguments passthrough to qemu as -k v |
| power-management | i | specify the power management passthrough mode |
| - 1 : pass-through mode (limited scope) | ||
| - 2 : non pass-through mode (in doubt use this) | ||
| oem-features | i | specify whether or not to pass through oem features. |
| Note: At the moment any integer value can be passed | ||
| but this is likely to change in future especially if | ||
| we decide to pass through a subset of oem features | ||
| and let user configure that subset. | ||
| timer-mode | i | specify the timer mode used. |
| timeoffset | s | specify the time offset (i.e. timezone) used. |
| pci-msitranslate | i | specify whether to use MSI-INTx translation for guest. |
| pci-power-management | i | specify whether or not to enable Dx power management |
| for passthrough devices. | ||
| inject-sci | i | specify whether or not to inject SCIs like lid close, |
| power button press to guest. (Default: no injection) |
Sending command to the monitor
xenvm bind a unix socket to the uuid specified/generated You can easily send command to the monitor with xenvm-cmd using the simple syntax:
xenvm-cmd
The following command are supported:
- help
- pause
- unpause
- destroy
- start
- suspend file=
- suspend file= live=true
- restore file=
- checkpoint
- shutdown
- restart
- quit (quit the monitor leaving the vm untouched)
Running with xen-unstable
It's possible to run with xen-unstable directly, but you need the qemu-dm-wrapper script and xenguest binary in /usr/bin/. They are available in xenguest/xenguest and scripts/qemu-dm-wrapper in the toolstack.git repository.
And you need to replace your udev rules by the one available in scripts/xen-backend.rules and scripts/xen-frontend.rules and add the scripts/tap scripts/block scripts/block-front scripts/vif into /etc/xensource/scripts/
Also note that since xen-unstable doesn't have the dm-ready patch, hvm domain takes unfortunately a substantial time (around 20s.) to start.
Example of config
PV config with one LVM disk called 'test' and one VIF :
| Code Block | ||
|---|---|---|
| ||
uuid = 00000000-0000-0000-0000-000000000001 hvm = false kernel = /boot/vmlinuz-2.6.18-xenU cmdline = root=/dev/sda1 ro memory = 64 disk = /dev/vg/test:phy:sda:w:disk vif = |
HVM config for installing a windows 2k3 from iso on a LVM disk called 'testhvm':
| Code Block | ||
|---|---|---|
| ||
uuid = 00000000-0000-0000-0000-000000000002 hvm = true memory = 256 disk = /dev/vg/testhvm:phy:hda:w:disk disk = /var/opt/xen/iso_import/w2k3eesp2.iso:file:hdd:r:cdrom boot = dc pae = trueacpi = true apic = true |