KVM host & libvirt

require

grep -E --color=always 'vmx|svm| lm ' /proc/cpuinfo | uniq

install

apt install \
    bridge-utils \
    cpu-checker \
    qemu-kvm \
    qemu-system-x86
    # qemu-system
    # --no-install-recommends

lsmod | grep kvm
#rmmod kvm
#modprobe -a kvm

kvm-ok

ls -lF /dev/kvm
# crw-rw---- 1 root kvm 10, 232 May 10 05:30 /dev/kvm

[optional] libvirt

apt install \
    libguestfs-tools \
    libosinfo-bin \
    libvirt-clients \
    libvirt-daemon-system \
    virtinst

make virsh commands available to some non-root user

usermod -aG libvirt USER

ls -lF /var/run/libvirt/libvirt-sock # root:libvirt

enable the daemon

systemctl status libvirtd
systemctl status libvirt-guests

check

ps auxww | grep libvirtd$
virsh list --all
virsh nodeinfo

virt-host-validate
osinfo-query os | less
libguestfs-test-tool

nested

to enable nested virtualization e.g. to setup XEN dom0 within KVM

vi /etc/modprobe.d/kvm.conf

options kvm_amd nested=1

# doesn't show up
#options kvm_amd amd_iommu=pt

rmmod kvm_amd
modprobe kvm_amd
    modinfo kvm_amd | grep nested

cat /sys/module/kvm_amd/parameters/nested
#cat /sys/module/kvm_amd/parameters/amd_iommu

network

see host-network

additional notes

storage

assuming /data/ is an NFS mount point already, enable it as storage for vdisks and configs

virsh pool-list
virsh pool-dumpxml default > pool.xml
virsh pool-destroy default
vi pool.xml

    <path>/data/guests</path>

mkdir -p /data/guests/
virsh pool-create pool.xml
virsh pool-list

ready to go

mkdir -p /data/guests/

ifconfig virbr0

and install some guests

live-migration

to be able to talk to the other hypervisors and enable the farm, you will have to share a few SSH keys.

to migrate a guest, make sure the machine type is compatible on both ends,

kvm -M ?

and fix it (DO NOT USE the pc alias that is pointing to different latest versions depending on the hypervisors – choose the latest available on all hypervisors e.g. pc-i440fx-2.5) if required,

virsh shutdown GUESTNAME
virsh edit GUESTNAME

  <os>
    <type arch='x86_64' machine='pc-i440fx-2.5'>hvm</type>
    <boot dev='hd'/>
  </os>

the available virtual network devices can be displayed as such,

qemu-system-x86_64 -net nic,model=? /dev/null

e.g.,

qemu: Supported NIC models: ne2k_pci,i82551,i82557b,i82559er,rtl8139,e1000,pcnet,virtio

replacing rtl8139 with virtio,

    <interface type='bridge'>
      <mac address='xx:xx:xx:xx:xx:xx'/>
      <source bridge='xenbr0'/>
      <model type='virtio'/>
            <driver name='vhost' txmode='iothread' ioeventfd='on' event_idx='off' queues='5' rx_queue_size='256' tx_queue_size='256'>
              <host csum='off' gso='off' tso4='off' tso6='off' ecn='off' ufo='off' mrg_rxbuf='off'/>
              <guest csum='off' tso4='off' tso6='off' ecn='off' ufo='off'/>
            </driver>
    </interface>

you may also have to disable caching on the virtual drive,

  <driver name='qemu' type='qcow2' cache='none'/>

refs https://access.redhat.com/solutions/158363

you will then be able to live-migrate,

virsh -c qemu+ssh://ANOTHER-KVM-SRV/system list
virsh migrate GUESTNAME qemu+ssh://ANOTHER-KVM-SRV/system

Resources

KVM host & libvirt

require

install

[optional] libvirt

nested

network

additional notes

storage

ready to go

live-migration

Resources

install

tuning

virt-install

nested

kvm virtio

network

storage

live-migration