Control Groups Resource Management

The QEMU and LXC drivers make use of the Linux "Control Groups" facility for applying resource management to their virtual machines and containers.

Required controllers

The control groups filesystem supports multiple "controllers". By default the init system (such as systemd) should mount all controllers compiled into the kernel at /sys/fs/cgroup/$CONTROLLER-NAME. Libvirt will never attempt to mount any controllers itself, merely detect where they are mounted.

The QEMU driver is capable of using the cpuset, cpu, cpuacct, memory, blkio and devices controllers. None of them are compulsory. If any controller is not mounted, the resource management APIs which use it will cease to operate. It is possible to explicitly turn off use of a controller, even when mounted, via the /etc/libvirt/qemu.conf configuration file.

The LXC driver is capable of using the cpuset, cpu, cpuacct, freezer, memory, blkio and devices controllers. The cpuacct, devices and memory controllers are compulsory. Without them mounted, no containers can be started. If any of the other controllers are not mounted, the resource management APIs which use them will cease to operate.

Current cgroups layout

As of libvirt 1.0.5 or later, the cgroups layout created by libvirt has been simplified, in order to facilitate the setup of resource control policies by administrators / management applications. The new layout is based on the concepts of "partitions" and "consumers". A "consumer" is a cgroup which holds the processes for a single virtual machine or container. A "partition" is a cgroup which does not contain any processes, but can have resource controls applied. A "partition" will have zero or more child directories which may be either "consumer" or "partition".

As of libvirt 1.1.1 or later, the cgroups layout will have some slight differences when running on a host with systemd 205 or later. The overall tree structure is the same, but there are some differences in the naming conventions for the cgroup directories. Thus the following docs split in two, one describing systemd hosts and the other non-systemd hosts.

Systemd cgroups integration

On hosts which use systemd, each consumer maps to a systemd scope unit, while partitions map to a system slice unit.

Systemd scope naming

The systemd convention is for the scope name of virtual machines / containers to be of the general format machine-$NAME.scope. Libvirt forms the $NAME part of this by concatenating the driver type with the id and truncated name of the guest, and then escaping any systemd reserved characters. So for a guest demo running under the lxc driver, we get a $NAME of lxc-12345-demo which when escaped is lxc\x2d12345\x2ddemo. So the complete scope name is machine-lxc\x2d12345\x2ddemo.scope. The scope names map directly to the cgroup directory names.

Systemd slice naming

The systemd convention for slice naming is that a slice should include the name of all of its parents prepended on its own name. So for a libvirt partition /machine/engineering/testing, the slice name will be machine-engineering-testing.slice. Again the slice names map directly to the cgroup directory names. Systemd creates three top level slices by default, system.slice user.slice and machine.slice. All virtual machines or containers created by libvirt will be associated with machine.slice by default.

Systemd cgroup layout

Given this, a possible systemd cgroups layout involving 3 qemu guests, 3 lxc containers and 3 custom child slices, would be: 

$ROOT
  |
  +- system.slice
  |   |
  |   +- libvirtd.service
  |
  +- machine.slice
      |
      +- machine-qemu\x2d1\x2dvm1.scope
      |   |
      |   +- emulator
      |   +- vcpu0
      |   +- vcpu1
      |
      +- machine-qemu\x2d2\x2dvm2.scope
      |   |
      |   +- emulator
      |   +- vcpu0
      |   +- vcpu1
      |
      +- machine-qemu\x2d3\x2dvm3.scope
      |   |
      |   +- emulator
      |   +- vcpu0
      |   +- vcpu1
      |
      +- machine-engineering.slice
      |   |
      |   +- machine-engineering-testing.slice
      |   |   |
      |   |   +- machine-lxc\x2d11111\x2dcontainer1.scope
      |   |
      |   +- machine-engineering-production.slice
      |       |
      |       +- machine-lxc\x2d22222\x2dcontainer2.scope
      |
      +- machine-marketing.slice
          |
          +- machine-lxc\x2d33333\x2dcontainer3.scope

Non-systemd cgroups layout

On hosts which do not use systemd, each consumer has a corresponding cgroup named $VMNAME.libvirt-{qemu,lxc}. Each consumer is associated with exactly one partition, which also have a corresponding cgroup usually named $PARTNAME.partition. The exceptions to this naming rule are the three top level default partitions, named /system (for system services), /user (for user login sessions) and /machine (for virtual machines and containers). By default every consumer will of course be associated with the /machine partition.

Given this, a possible systemd cgroups layout involving 3 qemu guests, 3 lxc containers and 2 custom child slices, would be: 

$ROOT
  |
  +- system
  |   |
  |   +- libvirtd.service
  |
  +- machine
      |
      +- qemu-1-vm1.libvirt-qemu
      |   |
      |   +- emulator
      |   +- vcpu0
      |   +- vcpu1
      |
      +- qeme-2-vm2.libvirt-qemu
      |   |
      |   +- emulator
      |   +- vcpu0
      |   +- vcpu1
      |
      +- qemu-3-vm3.libvirt-qemu
      |   |
      |   +- emulator
      |   +- vcpu0
      |   +- vcpu1
      |
      +- engineering.partition
      |   |
      |   +- testing.partition
      |   |   |
      |   |   +- lxc-11111-container1.libvirt-lxc
      |   |
      |   +- production.partition
      |       |
      |       +- lxc-22222-container2.libvirt-lxc
      |
      +- marketing.partition
          |
          +- lxc-33333-container3.libvirt-lxc

Using custom partitions

If there is a need to apply resource constraints to groups of virtual machines or containers, then the single default partition /machine may not be sufficiently flexible. The administrator may wish to sub-divide the default partition, for example into "testing" and "production" partitions, and then assign each guest to a specific sub-partition. This is achieved via a small element addition to the guest domain XML config, just below the main domain element 

...
<resource>
  <partition>/machine/production</partition>
</resource>
...

Note that the partition names in the guest XML are using a generic naming format, not the low level naming convention required by the underlying host OS. That is, you should not include any of the .partition or .slice suffixes in the XML config. Given a partition name /machine/production, libvirt will automatically apply the platform specific translation required to get /machine/production.partition (non-systemd) or /machine.slice/machine-production.slice (systemd) as the underlying cgroup name

Libvirt will not auto-create the cgroups directory to back this partition. In the future, libvirt / virsh will provide APIs / commands to create custom partitions, but currently this is left as an exercise for the administrator.

Note: the ability to place guests in custom partitions is only available with libvirt >= 1.0.5, using the new cgroup layout. The legacy cgroups layout described later in this document did not support customization per guest.

Creating custom partitions (systemd)

Given the XML config above, the admin on a systemd based host would need to create a unit file /etc/systemd/system/machine-production.slice 

# cat > /etc/systemd/system/machine-testing.slice <<EOF
[Unit]
Description=VM testing slice
Before=slices.target
Wants=machine.slice
EOF
# systemctl start machine-testing.slice

Creating custom partitions (non-systemd)

Given the XML config above, the admin on a non-systemd based host would need to create a cgroup named '/machine/production.partition' 

# cd /sys/fs/cgroup
# for i in blkio cpu,cpuacct cpuset devices freezer memory net_cls perf_event
  do
    mkdir $i/machine/production.partition
  done
# for i in cpuset.cpus  cpuset.mems
  do
    cat cpuset/machine/$i > cpuset/machine/production.partition/$i
  done

Resource management APIs/commands

Since libvirt aims to provide an API which is portable across hypervisors, the concept of cgroups is not exposed directly in the API or XML configuration. It is considered to be an internal implementation detail. Instead libvirt provides a set of APIs for applying resource controls, which are then mapped to corresponding cgroup tunables

Scheduler tuning

Parameters from the "cpu" controller are exposed via the schedinfo command in virsh. 

# virsh schedinfo demo
Scheduler      : posix
cpu_shares     : 1024
vcpu_period    : 100000
vcpu_quota     : -1
emulator_period: 100000
emulator_quota : -1

Block I/O tuning

Parameters from the "blkio" controller are exposed via the bkliotune command in virsh. 

# virsh blkiotune demo
weight         : 500
device_weight  :

Memory tuning

Parameters from the "memory" controller are exposed via the memtune command in virsh. 

# virsh memtune demo
hard_limit     : 580192
soft_limit     : unlimited
swap_hard_limit: unlimited

Network tuning

The net_cls is not currently used. Instead traffic filter policies are set directly against individual virtual network interfaces.

Legacy cgroups layout

Prior to libvirt 1.0.5, the cgroups layout created by libvirt was different from that described above, and did not allow for administrator customization. Libvirt used a fixed, 3-level hierarchy libvirt/{qemu,lxc}/$VMNAME which was rooted at the point in the hierarchy where libvirtd itself was located. So if libvirtd was placed at /system/libvirtd.service by systemd, the groups for each virtual machine / container would be located at /system/libvirtd.service/libvirt/{qemu,lxc}/$VMNAME. In addition to this, the QEMU drivers further child groups for each vCPU thread and the emulator thread(s). This leads to a hierarchy that looked like 

$ROOT
  |
  +- system
      |
      +- libvirtd.service
           |
           +- libvirt
               |
               +- qemu
               |   |
               |   +- vm1
               |   |   |
               |   |   +- emulator
               |   |   +- vcpu0
               |   |   +- vcpu1
               |   |
               |   +- vm2
               |   |   |
               |   |   +- emulator
               |   |   +- vcpu0
               |   |   +- vcpu1
               |   |
               |   +- vm3
               |       |
               |       +- emulator
               |       +- vcpu0
               |       +- vcpu1
               |
               +- lxc
                   |
                   +- container1
                   |
                   +- container2
                   |
                   +- container3

Although current releases are much improved, historically the use of deep hierarchies has had a significant negative impact on the kernel scalability. The legacy libvirt cgroups layout highlighted these problems, to the detriment of the performance of virtual machines and containers.