Supporting Hardware - Expert Consolidation in Oracle Database 12c - page 86

memory /cgroup/memory
devices /cgroup/devices
freezer /cgroup/freezer
net_cls /cgroup/net_cls
blkio /cgroup/blkio

■ this section is about the cpuset subsystem only. the complete reference to Control groups can be found in the
oracle Linux administrator's Solution guide for release 6 in Chapter 8. You are encouraged to review it to get a better
understanding about all the interesting features available in the other systems!

Note

The preceding output shows the default configuration from a recent Oracle Linux 6 system with package
libcgroup installed. If enabled, the cgconfig service, which is part of the package, will read the configuration file in
/etc/cgconfig.conf and mount the subsystems accordingly. Interestingly, the control group hierarchy is not exposed
in the output of the mount command but rather in /proc/mounts .
From a consolidation point of view, it makes sense to limit processes to CPU socket(s). The aim of this section
is to create two databases and bind all their respective processes to an individual socket and its memory. Control
Groups are externalized in userland similar to the sys file system sysfs in Linux by a set of virtual files. Figure 3-3
demonstrates the relationship between the different subsystems and Control Groups.

Figure 3-3. The cgroup example explained

Thankfully for the administrator, the first step of defining mount points for each subsystem is already done for us
with the installation of the libcgroup package. You read earlier that the cgroup mount point resembles the SYSFS mount,
following the UNIX paradigm that everything is a file. The contents of the top level cpuset directory are shown here:

[root@server1 ~]# ls /cgroup/cpuset/
cgroup.clone_children cpuset.memory_migrate cpuset.sched_relax_domain_level
cgroup.event_control cpuset.memory_pressure notify_on_release