github.com/opencontainers/runc@v1.2.0-rc.1.0.20240520010911-492dc558cdd6/docs/cgroup-v2.md (about) 1 # cgroup v2 2 3 runc fully supports cgroup v2 (unified mode) since v1.0.0-rc93. 4 5 To use cgroup v2, you might need to change the configuration of the host init system. 6 The following distributions are known to use cgroup v2 by default: 7 <!-- the list should be kept in sync with https://github.com/rootless-containers/rootlesscontaine.rs/blob/master/content/getting-started/common/cgroup2.md --> 8 - Fedora (since 31) 9 - Arch Linux (since April 2021) 10 - openSUSE Tumbleweed (since c. 2021) 11 - Debian GNU/Linux (since 11) 12 - Ubuntu (since 21.10) 13 - RHEL and RHEL-like distributions (since 9) 14 15 On other systemd-based distros, cgroup v2 can be enabled by adding `systemd.unified_cgroup_hierarchy=1` to the kernel cmdline. 16 17 ## Am I using cgroup v2? 18 19 Yes if `/sys/fs/cgroup/cgroup.controllers` is present. 20 21 ## Host Requirements 22 ### Kernel 23 * Recommended version: 5.2 or later 24 * Minimum version: 4.15 25 26 Kernel older than 5.2 is not recommended due to lack of freezer. 27 28 Notably, kernel older than 4.15 MUST NOT be used (unless you are running containers with user namespaces), as it lacks support for controlling permissions of devices. 29 30 ### Systemd 31 On cgroup v2 hosts, it is highly recommended to run runc with the systemd cgroup driver (`runc --systemd-cgroup`), though not mandatory. 32 33 The recommended systemd version is 244 or later. Older systemd does not support delegation of `cpuset` controller. 34 35 Make sure you also have the `dbus-user-session` (Debian/Ubuntu) or `dbus-daemon` (CentOS/Fedora) package installed, and that `dbus` is running. On Debian-flavored distros, this can be accomplished like so: 36 37 ```console 38 $ sudo apt install -y dbus-user-session 39 $ systemctl --user start dbus 40 ``` 41 42 ## Rootless 43 On cgroup v2 hosts, rootless runc can talk to systemd to get cgroup permissions to be delegated. 44 45 ```console 46 $ runc spec --rootless 47 $ jq '.linux.cgroupsPath="user.slice:runc:foo"' config.json | sponge config.json 48 $ runc --systemd-cgroup run foo 49 ``` 50 51 The container processes are executed in a cgroup like `/user.slice/user-$(id -u).slice/user@$(id -u).service/user.slice/runc-foo.scope`. 52 53 ### Configuring delegation 54 Typically, only `memory` and `pids` controllers are delegated to non-root users by default. 55 56 ```console 57 $ cat /sys/fs/cgroup/user.slice/user-$(id -u).slice/user@$(id -u).service/cgroup.controllers 58 memory pids 59 ``` 60 61 To allow delegation of other controllers, you need to change the systemd configuration as follows: 62 63 ```console 64 # mkdir -p /etc/systemd/system/user@.service.d 65 # cat > /etc/systemd/system/user@.service.d/delegate.conf << EOF 66 [Service] 67 Delegate=cpu cpuset io memory pids 68 EOF 69 # systemctl daemon-reload 70 ```