trentdocs_website/docs/stupid_kvm_tricks.md

229 lines
5.9 KiB
Markdown
Raw Permalink Normal View History

2018-02-10 23:31:16 -08:00
# Stupid KVM Tricks
## virt-install ubuntu16.04
Create the disk image
`qemu-img create -f qcow2 /var/lib/libvirt/images/xenial.qcow2 20G`
Command to run the install
```bash
virt-install \
--name xenial \
--ram 4096 \
--disk path=/var/lib/libvirt/images/xenial.qcow2,size=20 \
--vcpus 4 \
--os-type linux \
--os-variant ubuntu16.04 \
--network bridge=br0 \
--graphics none \
--console pty,target_type=serial \
--location ./ubuntu-16.04.3-server-amd64.iso \
--extra-args 'console=ttyS0,115200n8 serial'
```
## virt-install Arch Linux
The `--extra-args` option lets you use a serial console. But the
`--extra-args` option only works if you also use an `--location`
option. But the `--location` option can only be used with certain isos.
So use `--cdrom` instead of `--location`, drop the `--extra-args`,
and instruct the kernel to boot with a serial console with a parameter
at the boot splash screen.
2018-02-10 23:31:16 -08:00
```bash
qemu-img create -f qcow2 /var/lib/libvirt/images/arch.qcow2 20G
virt-install --name arch --ram 4096 \
--disk path=/var/lib/libvirt/images/arch.qcow2,size=20 \
--vcpus 2 \
--os-type linux \
--os-variant ubuntu16.04 \
--network bridge=virbr0 \
--graphics none \
--console pty,target_type=serial \
--cdrom /var/lib/libvirt/images/archlinux-2018.02.01-x86_64.iso
```
the arch boot splash screen will appear in your terminal and you can
tap the "tab" key to edit boot parameters
add "console=ttyS0" to kernel command line parameters
before
```bash
> .linux boot/x86_64/vmlinuz archisobasedir=arch archisolabel=ARCH_201802 initrd=boot/intel_ucode.img,boot/x86_64/archiso.img
```
after
```bash
> .linux boot/x86_64/vmlinuz archisobasedir=arch archisolabel=ARCH_201802 initrd=boot/intel_ucode.img,boot/x86_64/archiso.img console=ttyS0
```
```bash
arch boots ...
...
...
...
root@archiso ~ # lsblk
NAME MAJ:MIN RM SIZE RO TYPE MOUNTPOINT
loop0 7:0 0 432M 1 loop /run/archiso/sfs/airootfs
sr0 11:0 1 539M 0 rom /run/archiso/bootmnt
vda 254:0 0 20G 0 disk
root@archiso ~ #
```
2018-02-10 23:31:16 -08:00
## Change the Network Interface
br0 gets addresses from the network router, but what if you want
your vm to have be on the virbr0 192.168.122.0/24 subnet?
`virsh edit xenial`
And then 'J' all the way down to the bottom, change the interface name from br0 to
virbr0,
`virsh start xenial`
and then look for the machine with nmap
`nmap -sn 192.168.122.0/24`
## Clone the VM
In this case we don't have to pre-allocate the disk image because virt-clone will do that
for us.
```bash
virt-clone --original xenial --name xenial-clone \
--file /var/lib/libvirt/images/xenial-clone.qcow2
```
## Clone the VM to another Machine
First dump the xml that defines the virtual machine.
```bash
virsh dumpxml xenial > xenial.xml
```
Then copy both `xenial.xml` and `xenial.qcow2` to the new host machine. On the new kvm
host you'll want to at least make sure your vm has the correct CPU architecture.
The command to get a list of supported kvm cpu architectures is:
```bash
virsh cpu-models <arch>
# i.e.
virsh cpu-models x86_64
```
After you edit `xenial.xml` and update the correct cpu architecture, mv `xenial.qcow2`
to `/var/lib/libvirt/images/`, clone it. `virt-clone` will handle generating new
mac addresses for the network interfaces.
```xml
<cpu mode='custom' match='exact'>
<model fallback='allow'>Haswell-noTSX</model>
</cpu>
# i.e. change to above to
<cpu mode='custom' match='exact'>
<model fallback='allow'>SandyBridge</model>
</cpu>
```
```bash
virt-clone --original-xml xenial.xml --name xenial-clone \
--file /var/lib/libvirt/images/xenial-clone.qcow2
```
## What is the os-type and os-variant type names?
`osinfo-query os`
## misc
* Start the vm `virsh start xenial`
* List all the vms `virsh list --all`
* Stop the vm `virsh destroy xenial`
* Delete the vm `virsh undefine xenial`
## virsh help
The `virsh help` command returns a long chart of help information. But each section has
a keyword.
Take for instance the command `virsh help monitor`. From this we
see the `domiflist` subcommand (among others). Unfortunately `domifaddr` doesn't seem to
work on the Ubuntu:16.04 host, but there are other ways to find the ip address of
a virtual machine.
So now if you want to see what host interface the vm `xenial` is attached to,
type.
```bash
virsh domiflist xenial
```
which returns:
```bash
Interface Type Source Model MAC
-------------------------------------------------------
vnet1 bridge virbr0 virtio 52:54:00:58:bf:75
```
So now we can find the address of virbr0 on the host machine.
```bash
ifconfig virbr0
```
which returns:
```bash
virbr0 Link encap:Ethernet HWaddr 52:54:00:38:87:38
inet addr:192.168.122.1 Bcast:192.168.122.255 Mask:255.255.255.0
UP BROADCAST RUNNING MULTICAST MTU:1500 Metric:1
RX packets:1351 errors:0 dropped:0 overruns:0 frame:0
TX packets:3037 errors:0 dropped:0 overruns:0 carrier:0
collisions:0 txqueuelen:1000
RX bytes:232346 (232.3 KB) TX bytes:502916 (502.9 KB)
```
and thus we know what subnet to scan with nmap to find the ip address of the vm
```bash
nmap -sn 192.168.122.0/24
```
2018-02-11 08:33:05 -08:00
## Snapshots
Create snapshot of vm `dcing`
```bash
virsh snapshot-create-as --domain dcing --name dcing-snap0
```
But you don't need to name your snapshots because they are listed by time.
```bash
virsh snapshot-create --domain dcing
```
List snapshots for vm `dcing`
```bash
virsh snapshot-list --domain dcing
Name Creation Time State
------------------------------------------------------------
1518366561 2018-02-11 08:29:21 -0800 shutoff
dcing-snap0 2018-02-11 08:22:57 -0800 shutoff
```
Revert dcing to snap0
```bash
virsh snapshot-revert --domain dcing --snapshotname dcing-snap0
```
Delete snapshot
```bash
virsh snapshot-delete --domain dcing --snapshotname dcing-snap0
```