| Balbir Singh | 846c7bb | 2007-10-18 23:39:44 -0700 | [diff] [blame^] | 1 | Control Groupstats is inspired by the discussion at |
| 2 | http://lkml.org/lkml/2007/4/11/187 and implements per cgroup statistics as |
| 3 | suggested by Andrew Morton in http://lkml.org/lkml/2007/4/11/263. |
| 4 | |
| 5 | Per cgroup statistics infrastructure re-uses code from the taskstats |
| 6 | interface. A new set of cgroup operations are registered with commands |
| 7 | and attributes specific to cgroups. It should be very easy to |
| 8 | extend per cgroup statistics, by adding members to the cgroupstats |
| 9 | structure. |
| 10 | |
| 11 | The current model for cgroupstats is a pull, a push model (to post |
| 12 | statistics on interesting events), should be very easy to add. Currently |
| 13 | user space requests for statistics by passing the cgroup path. |
| 14 | Statistics about the state of all the tasks in the cgroup is returned to |
| 15 | user space. |
| 16 | |
| 17 | NOTE: We currently rely on delay accounting for extracting information |
| 18 | about tasks blocked on I/O. If CONFIG_TASK_DELAY_ACCT is disabled, this |
| 19 | information will not be available. |
| 20 | |
| 21 | To extract cgroup statistics a utility very similar to getdelays.c |
| 22 | has been developed, the sample output of the utility is shown below |
| 23 | |
| 24 | ~/balbir/cgroupstats # ./getdelays -C "/cgroup/a" |
| 25 | sleeping 1, blocked 0, running 1, stopped 0, uninterruptible 0 |
| 26 | ~/balbir/cgroupstats # ./getdelays -C "/cgroup" |
| 27 | sleeping 155, blocked 0, running 1, stopped 0, uninterruptible 2 |