Managing Resources Using Control Groups

8 Managing Resources Using Control Groups

Control groups, referred to as cgroups, are an Oracle Linux kernel feature that organizes systemd services, and if required, individual processes (PIDs), into hierarchical groups for allocating system resources, such as CPU, memory, and I/O.

For example, if you have identified three processes that need to be allocated CPU time in a ratio of 150:100:50, you can create three cgroups, each with a CPU weight corresponding to one of the three values in the ratio, and assign the appropriate process to each cgroup.

Important:

Use systemd to configure cgroups.

Manual creation of cgroup directories in the /sys/fs/cgroup virtual file system (as discussed in this topic) can be helpful for illustrating underlying concepts. However, this approach should only be used for specific scenarios, such as temporary debugging or testing. For most use cases, we recommend using systemd to configure cgroups to ensure correct and persistent resource management.

By default, systemd creates a cgroup for the following:

Each systemd service set up on the host.

For example, a server might have control group NetworkManager.service to group processes owned by the NetworkManager service, and control group firewalld.service to group processes owned by the firewalld service, and so on.
Each user (UID) on the host.

The cgroup functionality is mounted as a virtual file system under /sys/fs/cgroup. Each cgroup has a corresponding directory within /sys/fs/cgroup file system. For example, the cgroups created by systemd for the services it manages can be seen by running the command ls -l /sys/fs/cgroup/system.slice | grep ".service" as shown in the following sample code block:

ls -l /sys/fs/cgroup/system.slice | grep ".service"
            ...root root 0 Mar 22 10:47 atd.service
            ...root root 0 Mar 22 10:47 auditd.service
            ...root root 0 Mar 22 10:47 chronyd.service
            ...root root 0 Mar 22 10:47 crond.service
            ...root root 0 Mar 22 10:47 dbus-broker.service
            ...root root 0 Mar 22 10:47 dtprobed.service
            ...root root 0 Mar 22 10:47 firewalld.service
            ...root root 0 Mar 22 10:47 httpd.service
            ...

You can also create custom cgroups by creating directories under the /sys/fs/cgroup virtual file system and assigning process IDs (PIDs) to different cgroups according to system requirements. However, the recommended practice is to use systemd to configure cgroups instead of creating the cgroups manually under /sys/fs/cgroup. See Oracle Linux 8: Managing the System With systemd for the recommended method of managing cgroups through systemd.

Oracle Linux provides two types of control groups:

Control groups version 1 (cgroups v1): These groups provide a per-resource controller hierarchy. Each resource, such as CPU, memory, I/O, and so on, has its own control group hierarchy. A disadvantage of this group is the difficulty of establishing proper coordination of resource use among groups that might belong to different process hierarchies.
Control groups version 2 (cgroups v2): These groups provide a single control group hierarchy against which all resource controllers are mounted. In this hierarchy, you can obtain better proper coordination of resource uses across different resource controllers. This version is an improvement over cgroups v1 whose over flexibility prevented proper coordination of resource use among the system consumers.

Both versions are present in Oracle Linux. However, by default, the cgroups v1 functionality is enabled and mounted on Oracle Linux 8 systems.

For more information about control groups, see the cgroups(7) and sysfs(5) manual pages.