码迷,mamicode.com
首页 > 其他好文 > 详细

cgroup--device systemd-cgls

时间:2020-11-21 12:35:14      阅读:7      评论:0      收藏:0      [点我收藏+]

标签:any   RoCE   ber   wing   cto   sar   nec   class   file   

 

systemd-cgls

 

2. How to use cgroups?
The user can access and manage cgroups directly and indirectly (with LXC, libvirt or Docker).
Install the necessary packages:

$ sudo apt-get install libcgroup1 cgroup-tools
Now, the enabled cgroups can be seen via proc filesystem or sysfs:

$ cat /proc/cgroups

#subsys_name    hierarchy       num_cgroups     enabled
cpuset  9       2       1
cpu     4       134     1
cpuacct 4       134     1
blkio   7       134     1
memory  5       163     1
devices 11      134     1
freezer 2       2       1
net_cls 3       2       1
perf_event      10      2       1
net_prio        3       2       1
hugetlb 8       2       1
pids    6       136     1

$ ls -l /sys/fs/cgroup/

total 0
dr-xr-xr-x 6 root root  0 Nov 13 00:55 blkio
drwxr-xr-x 2 root root 60 Nov 13 01:00 cgmanager
lrwxrwxrwx 1 root root 11 Nov 13 00:55 cpu -> cpu,cpuacct
lrwxrwxrwx 1 root root 11 Nov 13 00:55 cpuacct -> cpu,cpuacct
dr-xr-xr-x 6 root root  0 Nov 13 00:55 cpu,cpuacct
dr-xr-xr-x 3 root root  0 Nov 13 00:55 cpuset
dr-xr-xr-x 6 root root  0 Nov 13 00:55 devices
dr-xr-xr-x 3 root root  0 Nov 13 00:55 freezer
dr-xr-xr-x 3 root root  0 Nov 13 00:55 hugetlb
dr-xr-xr-x 6 root root  0 Nov 13 00:55 memory
lrwxrwxrwx 1 root root 16 Nov 13 00:55 net_cls -> net_cls,net_prio
dr-xr-xr-x 3 root root  0 Nov 13 00:55 net_cls,net_prio
lrwxrwxrwx 1 root root 16 Nov 13 00:55 net_prio -> net_cls,net_prio
dr-xr-xr-x 3 root root  0 Nov 13 00:55 perf_event
dr-xr-xr-x 6 root root  0 Nov 13 00:55 pids
dr-xr-xr-x 6 root root  0 Nov 13 00:55 systemd
cgroups can be configured directly via the sysfs. For example, let’s create a small bash script named test_cgroups.sh for demonstration:

#!/bin/bash

while :
do
    echo "Print line" > /dev/tty
    sleep 5
done
Run above script:

$ chmod +x test_cgroups.sh
$ ./test_cgroups.sh
Print line
Print line
Print line
...
...
Change directory to /sys/fs/cgroup/devices where devices represents kind of resources that allows or denies access to devices by tasks in a cgroup:

$ cd sys/fs/cgroup/devices
Then, create a directory cgroups_test_group:

# mkdir cgroups_test_group
After creation of the cgroups_test_group directory, the following files will be generated:

$ ls -l /sys/fs/cgroup/devices/cgroups_test_group

total 0
-rw-r--r-- 1 root root 0 Nov 16 02:05 cgroup.clone_children
-rw-r--r-- 1 root root 0 Nov 16 02:05 cgroup.procs
--w------- 1 root root 0 Nov 16 02:05 devices.allow
--w------- 1 root root 0 Nov 16 02:05 devices.deny
-r--r--r-- 1 root root 0 Nov 16 02:05 devices.list
-rw-r--r-- 1 root root 0 Nov 16 02:05 notify_on_release
-rw-r--r-- 1 root root 0 Nov 16 02:05 tasks
The tasks file contains PIDs (Process ID) of processes which will be attached to the cgroups_test_group, the devices.deny file contains list of denied devices. By default, a newly created group has no any limits for devices access. In order to forbid a device (in this case, it’s /dev/tty), the devices.deny file should be modified:

# echo "c 5:0 w" > devices.deny
In the above command, the c indicates that /dev/tty is a character device, 5:0 is major and minor numbers of the device. The last w is write permission, so the above command forbids tasks to write to the /dev/tty.

$ ls -l /dev/tty

crw-rw-rw- 1 root tty 5, 0 Nov 18 17:02 /dev/tty
After that, re-run the script test_cgroups.sh:

$ ./test_cgroups.sh
Print line
Print line
Print line
...
...
then add the PID of this process to the tasks file:

# echo $(pidof -x test_cgroups.sh) > /sys/fs/cgroup/devices/cgroups_test_group/tasks
The result will be as expected:

$ ./test_cgroups.sh
Print line
Print line
Print line
./test_cgroups.sh: line 5: /dev/tty: Operation not permitted
./test_cgroups.sh: line 5: /dev/tty: Operation not permitted
...
...
An other example when running docker container

$ docker ps
CONTAINER ID        IMAGE               COMMAND             CREATED             STATUS              PORTS
98225055fa39        ubuntu              "/bin/bash"         47 seconds ago      Up 30 seconds

$ cat /sys/fs/cgroup/device/docker/98225055fa394b388e988b067b77dda61e53027ee944e4e0fd7887e19cdcf341/tasks
13556
During starting up of a docker container, docker creates a cgroup for processes in this container:

$ docker run -it ubuntu
$ top
  PID USER      PR  NI    VIRT    RES    SHR S  %CPU %MEM     TIME+ COMMAND
   1  root      20   0   18508   1848   1444 S   0.0  0.0   0:00.01 bash
   12 root      20   0   36628   1924   1420 R   0.0  0.0   0:00.01 top
Now, the cgroup of above process will be seen on host machine:

$ systemd-cgls
Control group /:
-.slice
├─1429 /sbin/cgmanager -m name=systemd
├─docker
│ └─98225055fa394b388e988b067b77dda61e53027ee944e4e0fd7887e19cdcf341
│   └─13556 /bin/bash

 

 

 

type
type can have one of the following three values:
a — applies to all devices, both character devices and block devices
b — specifies a block device
c — specifies a character device

 

在/devices/cgroup  目录下创建目录  first,并设置禁止设备读:
root@ubuntu:/sys/fs/cgroup/devices# mkdir first
 
root@ubuntu:/sys/fs/cgroup/devices/first# echo "a 1:5 r" > devices.deny

在另外一个终端中设置:

root@ubuntu:~# cgexec -g devices:first dd if=/dev/zero of=zero bs=1M count=128 &
可见提示如下:
[1] 8973

 

root@ubuntu:/sys/fs/cgroup/devices# mkdir first
root@ubuntu:/sys/fs/cgroup/devices# ls -al
total 0
dr-xr-xr-x 10 root root   0 Sep 24 18:06 .
drwxr-xr-x 15 root root 380 Sep 24 18:06 ..
-rw-r--r--  1 root root   0 Sep 25 06:25 cgroup.clone_children
-rw-r--r--  1 root root   0 Sep 25 06:25 cgroup.procs
-r--r--r--  1 root root   0 Sep 25 06:25 cgroup.sane_behavior
drwxr-xr-x  2 root root   0 Oct 16 10:07 default
--w-------  1 root root   0 Sep 25 06:25 devices.allow
--w-------  1 root root   0 Sep 25 06:25 devices.deny
-r--r--r--  1 root root   0 Sep 25 06:25 devices.list
drwxr-xr-x  3 root root   0 Oct  9 15:45 docker
drwxr-xr-x  2 root root   0 Nov 17 19:47 first
drwxr-xr-x  4 root root   0 Oct 13 18:45 kubepods
drwxr-xr-x  4 root root   0 Oct 13 22:56 kubepods.slice
-rw-r--r--  1 root root   0 Sep 25 06:25 notify_on_release
-rw-r--r--  1 root root   0 Sep 25 06:25 release_agent
drwxr-xr-x 66 root root   0 Sep 24 18:06 system.slice
-rw-r--r--  1 root root   0 Sep 25 06:25 tasks
drwxr-xr-x  2 root root   0 Oct 31 11:10 test.slice
drwxr-xr-x  2 root root   0 Sep 24 18:06 user.slice
root@ubuntu:/sys/fs/cgroup/devices# ls first/
cgroup.clone_children  cgroup.procs  devices.allow  devices.deny  devices.list  notify_on_release  tasks
root@ubuntu:/sys/fs/cgroup/devices# 

 

cgroup--device systemd-cgls

标签:any   RoCE   ber   wing   cto   sar   nec   class   file   

原文地址:https://www.cnblogs.com/dream397/p/13996105.html

(0)
(0)
   
举报
评论 一句话评论(0
登录后才能评论!
© 2014 mamicode.com 版权所有  联系我们:gaon5@hotmail.com
迷上了代码!