1lxc(7)                                                                  lxc(7)
2
3
4

NAME

6       lxc - linux containers
7

OVERVIEW

9       The  container  technology is actively being pushed into the mainstream
10       Linux kernel. It provides resource management  through  control  groups
11       and resource isolation via namespaces.
12
13       lxc,  aims to use these new functionalities to provide a userspace con‐
14       tainer object which provides full resource isolation and resource  con‐
15       trol for an applications or a full system.
16
17       lxc  is  small  enough to easily manage a container with simple command
18       lines and complete enough to be used for other purposes.
19

REQUIREMENTS

21       The kernel version >= 3.10 shipped with the  distros,  will  work  with
22       lxc, this one will have less functionalities but enough to be interest‐
23       ing.
24
25       lxc relies on a set of functionalities  provided  by  the  kernel.  The
26       helper script lxc-checkconfig will give you information about your ker‐
27       nel configuration, required, and missing features.
28

FUNCTIONAL SPECIFICATION

30       A container is an object isolating some resources of the host, for  the
31       application or system running in it.
32
33       The  application / system will be launched inside a container specified
34       by a configuration that is either initially created or passed as a  pa‐
35       rameter of the commands.
36
37       How to run an application in a container
38
39       Before  running  an application, you should know what are the resources
40       you want to isolate. The default configuration is to isolate PIDs,  the
41       sysv  IPC  and mount points. If you want to run a simple shell inside a
42       container, a basic configuration is needed, especially if you  want  to
43       share  the  rootfs.  If  you  want to run an application like sshd, you
44       should provide a new network stack and a new hostname. If you  want  to
45       avoid conflicts with some files eg.  /var/run/httpd.pid, you should re‐
46       mount /var/run with an empty directory. If you want to avoid  the  con‐
47       flicts  in  all  the cases, you can specify a rootfs for the container.
48       The rootfs can be a directory tree, previously bind  mounted  with  the
49       initial rootfs, so you can still use your distro but with your own /etc
50       and /home
51
52       Here is an example of directory tree for sshd:
53
54
55       [root@lxc sshd]$ tree -d rootfs
56
57       rootfs
58       |-- bin
59       |-- dev
60       |   |-- pts
61       |   `-- shm
62       |       `-- network
63       |-- etc
64       |   `-- ssh
65       |-- lib
66       |-- proc
67       |-- root
68       |-- sbin
69       |-- sys
70       |-- usr
71       `-- var
72           |-- empty
73           |   `-- sshd
74           |-- lib
75           |   `-- empty
76           |       `-- sshd
77           `-- run
78               `-- sshd
79
80
81       and the mount points file associated with it:
82
83            [root@lxc sshd]$ cat fstab
84
85            /lib /home/root/sshd/rootfs/lib none ro,bind 0 0
86            /bin /home/root/sshd/rootfs/bin none ro,bind 0 0
87            /usr /home/root/sshd/rootfs/usr none ro,bind 0 0
88            /sbin /home/root/sshd/rootfs/sbin none ro,bind 0 0
89
90
91       How to run a system in a container
92
93       Running a system inside a container is paradoxically easier  than  run‐
94       ning  an application. Why? Because you don't have to care about the re‐
95       sources to be isolated, everything needs to be isolated, the other  re‐
96       sources  are  specified as being isolated but without configuration be‐
97       cause the container will set them up. eg.  the  ipv4  address  will  be
98       setup  by  the system container init scripts. Here is an example of the
99       mount points file:
100
101            [root@lxc debian]$ cat fstab
102
103            /dev /home/root/debian/rootfs/dev none bind 0 0
104            /dev/pts /home/root/debian/rootfs/dev/pts  none bind 0 0
105
106
107   CONTAINER LIFE CYCLE
108       When the container is created, it contains the  configuration  informa‐
109       tion.  When  a  process is launched, the container will be starting and
110       running. When the last process running inside the container exits,  the
111       container is stopped.
112
113       In  case  of  failure  when  the container is initialized, it will pass
114       through the aborting state.
115
116          ---------
117         | STOPPED |<---------------
118          ---------                 |
119              |                     |
120            start                   |
121              |                     |
122              V                     |
123          ----------                |
124         | STARTING |--error-       |
125          ----------         |      |
126              |              |      |
127              V              V      |
128          ---------    ----------   |
129         | RUNNING |  | ABORTING |  |
130          ---------    ----------   |
131              |              |      |
132         no process          |      |
133              |              |      |
134              V              |      |
135          ----------         |      |
136         | STOPPING |<-------       |
137          ----------                |
138              |                     |
139               ---------------------
140
141
142
143   CONFIGURATION
144       The container is configured through a configuration file, the format of
145       the configuration file is described in lxc.conf(5)
146
147   CREATING / DESTROYING CONTAINERS
148       A  persistent  container  object can be created via the lxc-create com‐
149       mand. It takes a container name as parameter and optional configuration
150       file  and template. The name is used by the different commands to refer
151       to this container. The lxc-destroy command will destroy  the  container
152       object.
153
154              lxc-create -n foo
155              lxc-destroy -n foo
156
157
158   VOLATILE CONTAINER
159       It  is  not  mandatory to create a container object before starting it.
160       The container can be directly started with a configuration file as  pa‐
161       rameter.
162
163   STARTING / STOPPING CONTAINER
164       When  the container has been created, it is ready to run an application
165       / system. This is the purpose of the  lxc-execute  and  lxc-start  com‐
166       mands.  If  the  container was not created before starting the applica‐
167       tion, the container will use the configuration file passed as parameter
168       to  the command, and if there is no such parameter either, then it will
169       use a default isolation. If the application ended, the  container  will
170       be  stopped, but if needed the lxc-stop command can be used to stop the
171       container.
172
173       Running an application inside a container is not exactly the same thing
174       as  running a system. For this reason, there are two different commands
175       to run an application into a container:
176
177              lxc-execute -n foo [-f config] /bin/bash
178              lxc-start -n foo [-f config] [/bin/bash]
179
180
181       The lxc-execute command will run the specified command into a container
182       via  an  intermediate process, lxc-init.  This lxc-init after launching
183       the specified command, will wait for its end and all  other  reparented
184       processes.  (to  support  daemons in the container). In other words, in
185       the container, lxc-init has PID 1 and the first process of the applica‐
186       tion has PID 2.
187
188       The  lxc-start  command  will directly run the specified command in the
189       container. The PID of the first process is 1. If no command  is  speci‐
190       fied  lxc-start  will run the command defined in lxc.init.cmd or if not
191       set, /sbin/init .
192
193       To summarize, lxc-execute is for running an application  and  lxc-start
194       is better suited for running a system.
195
196       If  the  application is no longer responding, is inaccessible or is not
197       able to finish by itself, a wild lxc-stop command  will  kill  all  the
198       processes in the container without pity.
199
200              lxc-stop -n foo -k
201
202
203   CONNECT TO AN AVAILABLE TTY
204       If  the  container is configured with ttys, it is possible to access it
205       through them. It is up to the container to provide a set  of  available
206       ttys  to  be used by the following command. When the tty is lost, it is
207       possible to reconnect to it without login again.
208
209              lxc-console -n foo -t 3
210
211
212   FREEZE / UNFREEZE CONTAINER
213       Sometime, it is useful to stop all the processes belonging  to  a  con‐
214       tainer, eg. for job scheduling. The commands:
215
216              lxc-freeze -n foo
217
218
219       will put all the processes in an uninteruptible state and
220
221              lxc-unfreeze -n foo
222
223
224       will resume them.
225
226       This  feature is enabled if the freezer cgroup v1 controller is enabled
227       in the kernel.
228
229   GETTING INFORMATION ABOUT CONTAINER
230       When there are a lot of containers, it is hard to follow what has  been
231       created or destroyed, what is running or what are the PIDs running in a
232       specific container. For this reason, the following commands may be use‐
233       ful:
234
235              lxc-ls -f
236              lxc-info -n foo
237
238
239       lxc-ls lists containers.
240
241       lxc-info gives information for a specific container.
242
243       Here  is an example on how the combination of these commands allows one
244       to list all the containers and retrieve their state.
245
246              for i in $(lxc-ls -1); do
247                lxc-info -n $i
248              done
249
250
251   MONITORING CONTAINER
252       It is sometime useful to track the states of a container,  for  example
253       to monitor it or just to wait for a specific state in a script.
254
255       lxc-monitor command will monitor one or several containers. The parame‐
256       ter of this command accepts a regular expression for example:
257
258              lxc-monitor -n "foo|bar"
259
260
261       will monitor the states of containers named 'foo' and 'bar', and:
262
263              lxc-monitor -n ".*"
264
265
266       will monitor all the containers.
267
268       For a container 'foo' starting, doing some work and exiting, the output
269       will be in the form:
270
271              'foo' changed state to [STARTING]
272              'foo' changed state to [RUNNING]
273              'foo' changed state to [STOPPING]
274              'foo' changed state to [STOPPED]
275
276
277       lxc-wait  command  will wait for a specific state change and exit. This
278       is useful for scripting to synchronize the launch of a container or the
279       end. The parameter is an ORed combination of different states. The fol‐
280       lowing example shows how to wait for a  container  if  it  successfully
281       started as a daemon.
282
283
284              # launch lxc-wait in background
285              lxc-wait -n foo -s STOPPED &
286              LXC_WAIT_PID=$!
287
288              # this command goes in background
289              lxc-execute -n foo mydaemon &
290
291              # block until the lxc-wait exits
292              # and lxc-wait exits when the container
293              # is STOPPED
294              wait $LXC_WAIT_PID
295              echo "'foo' is finished"
296
297
298
299   CGROUP SETTINGS FOR CONTAINERS
300       The  container  is  tied  with  the control groups, when a container is
301       started a control group is created and associated with it. The  control
302       group properties can be read and modified when the container is running
303       by using the lxc-cgroup command.
304
305       lxc-cgroup command is used to set or  get  a  control  group  subsystem
306       which  is associated with a container. The subsystem name is handled by
307       the user, the command won't do any syntax  checking  on  the  subsystem
308       name, if the subsystem name does not exists, the command will fail.
309
310              lxc-cgroup -n foo cpuset.cpus
311
312
313       will display the content of this subsystem.
314
315              lxc-cgroup -n foo cpu.shares 512
316
317
318       will set the subsystem to the specified value.
319

SEE ALSO

321       lxc(7),  lxc-create(1), lxc-copy(1), lxc-destroy(1), lxc-start(1), lxc-
322       stop(1), lxc-execute(1), lxc-console(1),  lxc-monitor(1),  lxc-wait(1),
323       lxc-cgroup(1),  lxc-ls(1), lxc-info(1), lxc-freeze(1), lxc-unfreeze(1),
324       lxc-attach(1), lxc.conf(5)
325

AUTHOR

327       Daniel Lezcano <daniel.lezcano@free.fr>
328
329       Christian Brauner <christian.brauner@ubuntu.com>
330
331       Serge Hallyn <serge@hallyn.com>
332
333       Stéphane Graber <stgraber@ubuntu.com>
334
335
336
337Version 4.0.10                    2021-09-18                            lxc(7)
Impressum