blob: eda40fd39cad9df7927e7409765007af38020219 [file] [log] [blame]
Balbir Singh846c7bb2007-10-18 23:39:44 -07001Control Groupstats is inspired by the discussion at
2http://lkml.org/lkml/2007/4/11/187 and implements per cgroup statistics as
3suggested by Andrew Morton in http://lkml.org/lkml/2007/4/11/263.
4
5Per cgroup statistics infrastructure re-uses code from the taskstats
6interface. A new set of cgroup operations are registered with commands
7and attributes specific to cgroups. It should be very easy to
8extend per cgroup statistics, by adding members to the cgroupstats
9structure.
10
11The current model for cgroupstats is a pull, a push model (to post
12statistics on interesting events), should be very easy to add. Currently
13user space requests for statistics by passing the cgroup path.
14Statistics about the state of all the tasks in the cgroup is returned to
15user space.
16
17NOTE: We currently rely on delay accounting for extracting information
18about tasks blocked on I/O. If CONFIG_TASK_DELAY_ACCT is disabled, this
19information will not be available.
20
21To extract cgroup statistics a utility very similar to getdelays.c
22has been developed, the sample output of the utility is shown below
23
24~/balbir/cgroupstats # ./getdelays -C "/cgroup/a"
25sleeping 1, blocked 0, running 1, stopped 0, uninterruptible 0
26~/balbir/cgroupstats # ./getdelays -C "/cgroup"
27sleeping 155, blocked 0, running 1, stopped 0, uninterruptible 2