From Funtoo
Revision as of 07:31, October 6, 2017 by Palica (talk | contribs) (Getting LXD)
Jump to: navigation, search

LXD is a container "hypervisor" it should provide user with a new and fresh experience using LXC technology.

LXD consists of three components:

  • A system-wide daemon (lxd)
  • A command line client (lxc)
  • An OpenStack Nova plugin (nova-compute-lxd)

A REST API that is accesible both locally and if enabled, over the network is provided from the lxd daemon.

The command line tool is designed to be a very simple, yet very powerful tool to manage all your containers. It can handle connections to multiple container hosts and easily give you an overview of all the containers on your network, let you create some more where you want them and even move them around while they're running.

The OpenStack plugin then allows you to use your lxd hosts as compute nodes, running workloads on containers rather than virtual machines.

The LXD project was founded and is currently led by Canonical Ltd and Ubuntu with contributions from a range of other companies and individual contributors.


Some of the biggest features of LXD are:

  • Secure by design (unprivileged containers, resource restrictions and much more)
  • Scalable (from containers on your laptop to thousand of compute nodes)
  • Intuitive (simple, clear API and crisp command line experience)
  • Image based (no more distribution templates, only good, trusted images)
  • Live migration

Unprivileged Containers

LXD uses unprivileged containers by default. The difference between an unprivileged container and a privileged one is whether the root user in the container is the “real” root user (uid 0 at the kernel level).

The way unprivileged containers are created is by taking a set of normal UIDs and GIDs from the host, usually at least 65536 of each (to be POSIX compliant) and mapping those into the container.

The most common example and what most LXD users will end up with by default is a map of 65536 UIDs and GIDs, with a host base id of 100000. This means that root in the container (uid 0) will be mapped to the host uid 100000 and uid 65535 in the container will be mapped to uid 165535 on the host. UID/GID 65536 and higher in the container aren’t mapped and will return an error if you attempt to use them.

From a security point of view, that means that anything which is not owned by the users and groups mapped into the container will be inaccessible. Any such resource will show up as being owned by uid/gid “-1” (rendered as 65534 or nobody/nogroup in userspace). It also means that should there be a way to escape the container, even root in the container would find itself with just as much privileges on the host as a nobody user.

LXD does offer a number of options related to unprivileged configuration:

  • Increasing the size of the default uid/gid map
  • Setting up per-container maps
  • Punching holes into the map to expose host users and groups

Relationship with LXC

LXD isn't a rewrite of LXC, in fact it's building on top of LXC to provide a new, better user experience. Under the hood, LXD uses LXC through liblxc and its Go binding to create and manage the containers.

It's basically an alternative to LXC's tools and distribution template system with the added features that come from being controllable over the network.


LXD is free software and is developed under the Apache 2 license.

Installing LXD in Funtoo

Kernel pre-requisities

These options should be disabled in your kernel to use all of the functions of LXD:


These options should be enabled in your kernel to use all of the functions of LXD:



The Funtoo's default kernel (sys-kernel/debian-sources – v. 4.11.11 at the time of writing) has all these options enabled.


On older kernels DEVPTS_MULTIPLE_INSTANCES is needed too (as of kernel version 4.11.11 - the option doesn't exist any more)

You can use this code to compare your config settings with the ones needed. Put the required config options in a kernel-req.txt file and run the script.

   kerncheck.py (python source code) - check kernel options
import gzip

REQF = "kernel-req.txt"    # copy kernel options requirements into this file
REQS = set()
CFGS = set()

with open(REQF) as f:
    for line in f:
        REQS.add("CONFIG_%s" % line.strip())

with gzip.open("/proc/config.gz") as f:
    for line in f:
        line = line.decode().strip()
        if not line or line.startswith("#"):

            [opt, val] = line.split("=")
            if val =="n":

print("Enabled config options:")
print(CFGS & REQS)

print("Missing config options:")
print(REQS - CFGS)

Getting LXD

Installing LXD is pretty straight forward as the ebuild exists in our portage tree. I would recommend putting /var on btrfs or zfs (or at least /var/lib/lxd) as LXD can take advantage of these COW filesytems. LXD doesn’t need any configuration to use btrfs, you just need to make sure that /var/lib/lxd is stored on a btrfs filesystem and LXD will automatically make use of it for you. You can use any other filesystem, but be advised LXD can take great advantage of btrfs or ZFS, be it for snapshots, clones, quotas and more. If you want to test it on your current filesystem consider creating a loop device that you format with btrfs and use that as your /var/lib/lxd device.

First we need to add some use flags to be able to emerge LXD. This is what is needed to have almost everything working.

app-emulation/lxc seccomp cgmanager
root # emerge -av lxd

These are the packages that would be merged, in order:

Calculating dependencies... done!
[ebuild  N     ] dev-lang/go-1.8-r1:0/1.8::gentoo  USE="-gccgo" 69,062 KiB
[ebuild  N     ] dev-go/go-crypto-0_pre20160126:0/0_pre20160126::gentoo  881 KiB
[ebuild  N     ] sys-fs/squashfs-tools-4.3-r2::gentoo  USE="xattr xz -debug -lz4 -lzma -lzo -static" 194 KiB
[ebuild  N     ] sys-libs/libseccomp-2.3.2::gentoo  USE="-static-libs" 547 KiB
[ebuild  N     ] net-libs/libnet-1.2_rc3-r1:1.1::gentoo  USE="-doc -static-libs" 661 KiB
[ebuild  N     ] dev-libs/libnl-3.3.0_rc1:3::gentoo  USE="python -static-libs -utils" PYTHON_TARGETS="python2_7 python3_4 -python3_5" 912 KiB
[ebuild  N     ] dev-python/ipaddr-2.1.11-r1::gentoo  PYTHON_TARGETS="python2_7 python3_4 -pypy -python3_5" 29 KiB
[ebuild  N     ] dev-go/go-text-0_pre20160211:0/0_pre20160211::gentoo  3,922 KiB
[ebuild  N     ] sys-libs/libcap-2.25::gentoo  USE="pam -static-libs" 63 KiB
[ebuild  N     ] dev-go/go-net-0_pre20160216:0/0_pre20160216::gentoo  724 KiB
[ebuild  N     ] net-dns/dnsmasq-2.76-r1::gentoo  USE="dhcp inotify ipv6 nls -auth-dns -conntrack -dbus -dhcp-tools -dnssec -idn -lua -script (-selinux) -static -tftp" LINGUAS="-de -es -fi -fr -id -it -no -pl -pt_BR -ro" 470 KiB
[ebuild  N     ] dev-libs/protobuf-c-1.2.1-r1:0/1.0.0::gentoo  USE="-static-libs {-test}" 448 KiB
[ebuild  N     ] sys-process/criu-2.12::gentoo  USE="python -setproctitle" PYTHON_TARGETS="python2_7" 632 KiB
[ebuild  N     ] app-emulation/lxc-2.0.7::gentoo  USE="python seccomp -cgmanager -doc -examples -lua" PYTHON_TARGETS="python3_4 -python3_5" 774 KiB
[ebuild  N     ] app-emulation/lxd-2.11::gentoo  USE="daemon nls {-test}" LINGUAS="-de -el -fr -ja -nl -ru" 2,352 KiB

Total: 15 packages (15 new), Size of downloads: 81,664 KiB

Would you like to add these changes to your config files? [Yes/No]

Running LXD

Once installed you need to start the LXD daemon. By running:

root # service lxd start
 * Starting lxd server ...

First setup of LXD/Initialisation

Before using LXD for the first time as a user, you may initialize your LXD environment. As recommended earlier I am using btrfs for this installation.

user $ lxd init
Do you want to configure a new storage pool (yes/no) [default=yes]? yes
Name of the new storage pool [default=default]: default
Name of the storage backend to use (dir, btrfs, lvm) [default=dir]: btrfs
Create a new BTRFS pool (yes/no) [default=yes]? yes
Would you like to use an existing block device (yes/no) [default=no]? no
Would you like to create a new subvolume for the BTRFS storage pool (yes/no) [default=yes]: yes
Would you like LXD to be available over the network (yes/no) [default=no]? no
Would you like stale cached images to be updated automatically (yes/no) [default=yes]? no
Would you like to create a new network bridge (yes/no) [default=yes]? yes
What should the new bridge be called [default=lxdbr0]? lxdbr0
What IPv4 address should be used (CIDR subnet notation, “auto” or “none”) [default=auto]? auto
What IPv6 address should be used (CIDR subnet notation, “auto” or “none”) [default=auto]? auto
LXD has been successfully configured.

What this does is it creates btrfs subvolumes like this:

user $ btrfs sub list .
ID 260 gen 1047 top level 5 path rootfs
ID 280 gen 1046 top level 260 path var/lib/lxd/storage-pools/default
ID 281 gen 1043 top level 280 path var/lib/lxd/storage-pools/default/containers
ID 282 gen 1044 top level 280 path var/lib/lxd/storage-pools/default/snapshots
ID 283 gen 1045 top level 280 path var/lib/lxd/storage-pools/default/images
ID 284 gen 1046 top level 280 path var/lib/lxd/storage-pools/default/custom

It also creates new network interface for you:

user $ ip a list dev lxdbr0
8: lxdbr0: <BROADCAST,MULTICAST,UP,LOWER_UP> mtu 1500 qdisc noqueue state UNKNOWN group default qlen 1000
    link/ether d2:9b:70:f2:8f:6f brd ff:ff:ff:ff:ff:ff
    inet scope global lxdbr0
       valid_lft forever preferred_lft forever
    inet brd scope global lxdbr0
       valid_lft forever preferred_lft forever
    inet6 fd42:efd8:662e:3184::1/64 scope global
       valid_lft forever preferred_lft forever
    inet6 fe80::caf5:b7ed:445e:b112/64 scope link
       valid_lft forever preferred_lft forever

And last but not least it also generates iptables rules for you:

user $ iptables -L
Chain INPUT (policy ACCEPT)
target     prot opt source               destination
ACCEPT     tcp  --  anywhere             anywhere             tcp dpt:domain /* generated for LXD network lxdbr0 */
ACCEPT     udp  --  anywhere             anywhere             udp dpt:domain /* generated for LXD network lxdbr0 */
ACCEPT     udp  --  anywhere             anywhere             udp dpt:bootps /* generated for LXD network lxdbr0 */

Chain FORWARD (policy ACCEPT)
target     prot opt source               destination
ACCEPT     all  --  anywhere             anywhere             /* generated for LXD network lxdbr0 */
ACCEPT     all  --  anywhere             anywhere             /* generated for LXD network lxdbr0 */

Chain OUTPUT (policy ACCEPT)
target     prot opt source               destination
ACCEPT     tcp  --  anywhere             anywhere             tcp spt:domain /* generated for LXD network lxdbr0 */
ACCEPT     udp  --  anywhere             anywhere             udp spt:domain /* generated for LXD network lxdbr0 */
ACCEPT     udp  --  anywhere             anywhere             udp spt:bootps /* generated for LXD network lxdbr0 */

user $ iptables -L -t nat
target     prot opt source               destination

Chain INPUT (policy ACCEPT)
target     prot opt source               destination

Chain OUTPUT (policy ACCEPT)
target     prot opt source               destination

target     prot opt source               destination
MASQUERADE  all  --     !      /* generated for LXD network lxdbr0 */

user $ iptables -L -t mangle
target     prot opt source               destination

Chain INPUT (policy ACCEPT)
target     prot opt source               destination

Chain FORWARD (policy ACCEPT)
target     prot opt source               destination

Chain OUTPUT (policy ACCEPT)
target     prot opt source               destination

target     prot opt source               destination
CHECKSUM   udp  --  anywhere             anywhere             udp dpt:bootpc /* generated for LXD network lxdbr0 */ CHECKSUM fill

Some other things done by the initialization and starting of the LXD daemon are:

  • dnsmasq listening on lxdbr0
  • ...

Finishing up the setup of LXD

There are still some things that you need to do manually. We need to setup subuid and subgid values for our containers to use. And for using non-systemd containers we will also need app-admin/cgmanager so emerge and start it now.

root # emerge app-admin/cgmanager
root # rc-update add lxd default
root # rc-update add cgmanager default
root # rc

Containers, snapshots and images

Containers in LXD are made of:

  • A filesystem (rootfs)
  • A list of configuration options, including resource limits, environment, security options and more
  • A bunch of devices like disks, character/block unix devices and network interfaces
  • A set of profiles the container inherits configuration from (see below)
  • Some properties (container architecture, ephemeral or persistent and the name)
  • Some runtime state (when using CRIU for checkpoint/restore)

Container snapshots as the name states snapshots of the container in time and cannot be modified in any way. It is worth noting that because snapshots can store the container runtime state, which gives us ability of “stateful” snapshots. That is, the ability to rollback the container including its cpu and memory state at the time of the snapshot.

LXD is image based, all LXD containers come from an image. Images are typically clean Linux distribution images similar to what you would use for a virtual machine or cloud instance. It is possible to “publish” a container, making an image from it which can then be used by the local or remote LXD hosts.

Our first image

Let's get our hands even more dirty and create our first image. We will be using a generic 64 bit Funtoo Linux image. Let's grab it, because we will need to modify it a little bit.

Modifying the stage3 image

So the steps that follow will download, extract the stage3 to a directory, and modify etc/rc.conf changing rc_sys value to lxc and comment out consoles in etc/inittab.

root # mkdir lxd-images
root # cd lxd-images
root # wget http://build.funtoo.org/funtoo-current/pure64/generic_64-pure64/stage3-latest.tar.xz
root # mkdir rootfs
root # tar -xaf stage3-latest.tar.xz -C rootfs/
root # cd rootfs
root # sed -i 's/^#rc_sys=""/rc_sys="lxc"/' etc/rc.conf
root # sed -i '/^c[1-6]/s/^\(.*\)$/#\1/' etc/inittab

Metadata and templates

Now we will create metadata and templates that will be used by the image and later by containers.

root # cd ..
root # mkdir templates
root # echo hostname=\"{{ container.name }}\" > templates/hostname.tpl

Create a file named metadata.yaml in current directory (lxd-images) with this contents:

architecture: x86_64
creation_date: 20170907
  architecture: x86_64
  description: Funtoo Current Generic Pure 64-bit
  name: funtoo-generic_64-pure64-funtoo-current-2016-12-10
  os: funtoo
  release: 1.0
  variant: current
    template: hostname.tpl
      - create
      - copy

Prepare the archive

Now we recreate the archive and later use it to import into LXD.

root # tar -caf lxd-image.tar.xz metadata.yaml templates rootfs

Import the image

After we have successfully recreated our archive we can now finally import it into LXD and start using it as our "seed" image for all our containers.

root # lxc image import lxd-image.tar.xz --alias funtoo
Image imported with fingerprint: e279c16d1a801b2bd1698df95e148e0a968846835f4769b24988f2eb3700100f
root # lxc image ls
| ALIAS  | FINGERPRINT  | PUBLIC |            DESCRIPTION             |  ARCH  |   SIZE   |         UPLOAD DATE         |
| funtoo | e279c16d1a80 | no     | Funtoo Current Generic Pure 64-bit | x86_64 | 347.75MB | Sep 8, 2017 at 1:17am (UTC) |

And there we have our very first Funtoo Linux image imported inside LXD. You can reference the image through the alias or through the fingerprint. Aliases can be added also later.

Let me show you some basic usage then.

Creating your first container

First we have to add some subuid and subgid values for lxd to use. Add these lines to your /etc/subuid and /etc/subgid files on the host.

root # nano -w /etc/subuid
root # nano -w /etc/subgid

The maps for “lxd” and “root” should always be kept in sync. LXD itself is restricted by the “root” allocation. The “lxd” entry is used to track what needs to be removed if LXD is uninstalled.

This isolates the users inside the container and even if they would escape the container they would get nobody's privileges on the host.

LXD daemon has to restart after making changes to subuid and subgid files.

root # service lxd restart

So now we can init our first container. That is done using this command:

root # lxc init funtoo c1
Creating c1
root #  lxc ls
| NAME |  STATE  | IPV4 | IPV6 |    TYPE    | SNAPSHOTS |
| c1   | STOPPED |      |      | PERSISTENT | 0         |

Profiles intermezzo

LXD has the ability to change quite a few container settings, including resource limitation, control of container startup and a variety of device pass-through options using what is called profiles. Let me show you how can this be used.

This is the default profile that gets inherited by all containers.

root # lxc profile list
|  NAME   | USED BY |
| default | 1       |

root #  lxc profile show default
config: {}
description: Default LXD profile
    nictype: bridged
    parent: lxdbr0
    type: nic
    path: /
    pool: default
    type: disk
name: default
- /1.0/containers/c1

Now let's make a profile specific for our funtoo containers. It will include shared meta-repo and some other goodies.

root # lxc profile create prf-funtoo
Profile prf-funtoo created
root #  lxc profile edit prf-funtoo
root ### This is a yaml representation of the profile.
root ### Any line starting with a '# will be ignored.
root ###
root ### A profile consists of a set of configuration items followed by a set of
root ### devices.
root ###
root ### An example would look like:
root ### name: onenic
root ### config:
root ###   raw.lxc: lxc.aa_profile=unconfined
root ### devices:
root ###   eth0:
root ###     nictype: bridged
root ###     parent: lxdbr0
root ###     type: nic
root ###
root ### Note that the name is shown but cannot be changed

  raw.lxc: lxc.mount.auto = proc sys cgroup
description: "LXD profile for Funtoo-based containers"
    path: var/git
    source: /var/git
    type: disk
name: prf-funtoo
used_by: []
root # lxc profile add c1 prf-funtoo
Profile prf-funtoo added to c1

Starting our first container

After we have done all these customizations we can now start our container.

root # lxc start c1

And now we can gain shell inside our container.

root # lxc exec c1 bash

Now you should see a different prompt starting with

c1 ~ #

If we run top or ps for example we will see only the processes of the container.

c1 ~ # ps aux
root         1  0.0  0.0   4248   748 ?        Ss+  13:20   0:00 init [3]
root       266  0.0  0.0  30488   472 ?        Ss   13:20   0:00 /usr/sbin/sshd
root       312  0.2  0.0  17996  3416 ?        Ss   13:29   0:00 bash
root       317  0.0  0.0  19200  2260 ?        R+   13:29   0:00 ps aux
c1 ~ #

As you can see only the container's processes are shown. User running the processes is root here. What happens if we search for all sshd processes for example on the host box?

root # ps aux|grep ssh
root     14505  0.0  0.0  30564  1508 ?        Ss   Sep07   0:00 /usr/sbin/sshd   
100000   25863  0.0  0.0  30488   472 ?        Ss   15:20   0:00 /usr/sbin/sshd   
root     29487  0.0  0.0   8324   828 pts/2    S+   15:30   0:00 grep --colour=auto sshd
root #

So as you can see, the sshd process is running under user with uid 100000 on the host machine and has a different PID.

Getting information about your containers

Listing containers

root #  lxc ls
| NAME |  STATE  |         IPV4         |                     IPV6                      |    TYPE    | SNAPSHOTS |
| c1   | RUNNING | (eth0) | fd42:156d:4593:a619:8619:546e:43f:2089 (eth0) | PERSISTENT | 0         |
|      |         |                      | fd42:156d:4593:a619:216:3eff:fe4a:3d4f (eth0) |            |           |

Container details

root # lxc info c1
Name: c1
Remote: unix://
Architecture: x86_64
Created: 2017/09/08 02:07 UTC
Status: Running
Type: persistent
Profiles: default, prf-funtoo
Pid: 6366
  eth0: inet   vethFG4HXG
  eth0: inet6   fd42:156d:4593:a619:8619:546e:43f:2089  vethFG4HXG
  eth0: inet6   fd42:156d:4593:a619:216:3eff:fe4a:3d4f  vethFG4HXG
  eth0: inet6   fe80::216:3eff:fe4a:3d4f        vethFG4HXG
  lo:   inet
  lo:   inet6   ::1
  Processes: 6
  CPU usage:
    CPU usage (in seconds): 25
  Memory usage:
    Memory (current): 69.01MB
    Memory (peak): 258.92MB
  Network usage:
      Bytes received: 83.65kB
      Bytes sent: 9.44kB
      Packets received: 188
      Packets sent: 93
      Bytes received: 0B
      Bytes sent: 0B
      Packets received: 0
      Packets sent: 0

Container configuration

root #  lxc config edit c1
root ### This is a yaml representation of the configuration.
root ### Any line starting with a '# will be ignored.
root ###
root ### A sample configuration looks like:
root ### name: container1
root ### profiles:
root ### - default
root ### config:
root ###   volatile.eth0.hwaddr: 00:16:3e:e9:f8:7f
root ### devices:
root ###   homedir:
root ###     path: /extra
root ###     source: /home/user
root ###     type: disk
root ### ephemeral: false
root ###
root ### Note that the name is shown but cannot be changed

architecture: x86_64
  image.architecture: x86_64
  image.description: Funtoo Current Generic Pure 64-bit
  image.name: funtoo-generic_64-pure64-funtoo-current-2016-12-10
  image.os: funtoo
  image.release: "1.0"
  image.variant: current
  volatile.base_image: e279c16d1a801b2bd1698df95e148e0a968846835f4769b24988f2eb3700100f
  volatile.eth0.hwaddr: 00:16:3e:4a:3d:4f
  volatile.eth0.name: eth0
  volatile.idmap.base: "0"
  volatile.idmap.next: '[{"Isuid":true,"Isgid":false,"Hostid":100000,"Nsid":0,"Maprange":65536},{"Isuid":false,"Isgid":true,"Hostid":100000,"Nsid":0,"Maprange":65536}]'
  volatile.last_state.idmap: '[{"Isuid":true,"Isgid":false,"Hostid":100000,"Nsid":0,"Maprange":65536},{"Isuid":false,"Isgid":true,"Hostid":100000,"Nsid":0,"Maprange":65536}]'
  volatile.last_state.power: RUNNING
devices: {}
ephemeral: false
- default
- prf-funtoo
stateful: false
description: ""

Managing files


Cloning, copying and moving containers

Resource control

LXD offers a variety of resource limits. Some of those are tied to the container itself, like memory quotas, CPU limits and I/O priorities. Some are tied to a particular device instead, like I/O bandwidth or disk usage limits.

As with all LXD configuration, resource limits can be dynamically changed while the container is running. Some may fail to apply, for example if setting a memory value smaller than the current memory usage, but LXD will try anyway and report back on failure.

All limits can also be inherited through profiles in which case each affected container will be constrained by that limit. That is, if you set limits.memory=256MB in the default profile, every container using the default profile (typically all of them) will have a memory limit of 256MB.


Setting a size limit on the container’s filesystem and have it enforced against the container. Right now LXD only supports disk limits if you’re using the ZFS or btrfs storage backend.

To set a disk limit (requires btrfs or ZFS):

root # lxc config device set c1 root size 20GB


To just limit a container to any 2 CPUs, do:

root # lxc config set c1 limits.cpu 2

To pin to specific CPU cores, say the second and fourth:

root # lxc config set c1 limits.cpu 1,3

More complex pinning ranges like this works too:

root # lxc config set c1 limits.cpu 0-3,7-11


To apply a straightforward memory limit run:

root # lxc config set c1 limits.memory 256MB

(The supported suffixes are kB, MB, GB, TB, PB and EB)

To turn swap off for the container (defaults to enabled):

root # lxc config set c1 limits.memory.swap false

To tell the kernel to swap this container’s memory first:

root # lxc config set c1 limits.memory.swap.priority 0

And finally if you don’t want hard memory limit enforcement:

root # lxc config set c1 limits.memory.enforce soft


Block I/O

Resource limits using profile - Funtoo Containers example

So I am going to create 3 profiles to mimic the resource limits for current Funtoo Containers.

PriceRAMCPU ThreadsDisk SpaceSign Up
$15/mo4GB6 CPU Threads50GBSign Up! (small)
$30/mo12GB12 CPU Threads100GBSign Up! (medium)
$45/mo48GB24 CPU Threads200GBSign Up! (large)

I am going to create one profile and copy/edit it for the remaining two options.

root # lxc profile create res-small
root # lxc profile edit res-small
  limits.cpu: "6"
  limits.memory: 4GB
description: Small Variant of Funtoo Containers
    path: /
    pool: default
    size: 50GB
    type: disk
name: small
used_by: []
root # lxc profile copy res-small res-medium
root # lxc profile copy res-small res-large
root # lxc profile set res-medium limits.cpu 12
root # lxc profile set res-medium limits.memory 12GB
root # lxc profile device set res-medium root size 100GB
root # lxc profile set res-large limits.cpu 24
root # lxc profile set res-large limits.memory 48GB
root # lxc profile device set res-large root size 200GB

Now let's create a container and assign the res-small and funtoo profiles to it.

root # lxc init funtoo c-small
root # lxc profile assign c-small res-small
root # lxc profile add c-small funtoo

Image manipulations

Remote hosts

Running systemd container on a non-systemd host

To use systemd in the container, a recent enough (>=4.6) kernel version with support for cgroup namespaces is needed. Additionally the host needs to have a name=systemd cgroup hierarchy mounted:

root # mkdir -p /sys/fs/cgroup/systemd
root # mount -t cgroup -o none,name=systemd systemd /sys/fs/cgroup/systemd

Doing so does not require running systemd on the host, it only allows to run systemd correctly inside the container(s) .

If you want to get systemd hierarchy mounted automatically on system startup, using /etc/fstab will not work, but the dev-libs/libcgroup can be used for this. First you needed to edit the /etc/cgroup/cgconfig.conf and add:

mount {
    "name=systemd" = /sys/fs/cgroup/systemd;

Then you need to start the cgconfig daemon:

root # rc-service cgconfig start

The daemon can be started as needed, or automatically at system start by simply adding it to default group:

root # rc-update add cgconfig default