1lxc-attach(1) lxc-attach(1)
2
6 lxc-attach - start a process inside a running container.
7
9 lxc-attach {-n, --name name} [-f, --rcfile config_file] [-a, --arch
10 arch] [-e, --elevated-privileges privileges] [-s,
11 --namespaces namespaces] [-R, --remount-sys-proc] [--keep-
12 env] [--clear-env] [-v, --set-var variable] [--keep-var
13 variable] [-- command]
14
16 lxc-attach runs the specified command inside the container specified by
17 name. The container has to be running already.
18 If no command is specified, the current default shell of the user run‐
20 ning lxc-attach will be looked up inside the container and executed.
21 This will fail if no such user exists inside the container or the con‐
22 tainer does not have a working nsswitch mechanism.
23 Previous versions of lxc-attach simply attached to the specified names‐
25 paces of a container and ran a shell or the specified command without
26 first allocating a pseudo terminal. This made them vulnerable to input
27 faking via a TIOCSTI ioctl call after switching between userspace exe‐
28 cution contexts with different privilege levels. Newer versions of lxc-
29 attach will try to allocate a pseudo terminal file descriptor pair on
30 the host and attach any standard file descriptors which refer to a ter‐
31 minal to the container side of the pseudo terminal before executing a
32 shell or command. Note, that if none of the standard file descriptors
33 refer to a terminal lxc-attach will not try to allocate a pseudo termi‐
34 nal. Instead it will simply attach to the containers namespaces and run
35 a shell or the specified command.
36
38 -f, --rcfile config_file
39 Specify the configuration file to configure the virtualization
40 and isolation functionalities for the container.
41 This configuration file if present will be used even if there is
43 already a configuration file present in the previously created
44 container (via lxc-create).
45 -a, --arch arch
47 Specify the architecture which the kernel should appear to be
48 running as to the command executed. This option will accept the
49 same settings as the lxc.arch option in container configuration
50 files, see lxc.conf(5). By default, the current architecture of
51 the running container will be used.
52 -e, --elevated-privileges privileges
54 Do not drop privileges when running command inside the contain‐
55 er. If this option is specified, the new process will not be
56 added to the container's cgroup(s) and it will not drop its ca‐
57 pabilities before executing.
58 You may specify privileges, in case you do not want to elevate
60 all of them, as a pipe-separated list, e.g. CGROUP|LSM. Allowed
61 values are CGROUP, CAP and LSM representing cgroup, capabilities
62 and restriction privileges respectively. (The pipe symbol needs
63 to be escaped, e.g. CGROUP\|LSM or quoted, e.g. "CGROUP|LSM".)
64 Warning: This may leak privileges into the container if the com‐
66 mand starts subprocesses that remain active after the main
67 process that was attached is terminated. The (re-)starting of
68 daemons inside the container is problematic, especially if the
69 daemon starts a lot of subprocesses such as cron or sshd. Use
70 with great care.
71 -s, --namespaces namespaces
73 Specify the namespaces to attach to, as a pipe-separated list,
74 e.g. NETWORK|IPC. Allowed values are MOUNT, PID, UTSNAME, IPC,
75 USER and NETWORK. This allows one to change the context of the
76 process to e.g. the network namespace of the container while re‐
77 taining the other namespaces as those of the host. (The pipe
78 symbol needs to be escaped, e.g. MOUNT\|PID or quoted, e.g.
79 "MOUNT|PID".)
80 Important: This option implies -e.
82 -R, --remount-sys-proc
84 When using -s and the mount namespace is not included, this flag
85 will cause lxc-attach to remount /proc and /sys to reflect the
86 current other namespace contexts.
87 Please see the Notes section for more details.
89 This option will be ignored if one tries to attach to the mount
91 namespace anyway.
92 --keep-env
94 Keep the current environment for attached programs. This is the
95 current default behaviour (as of version 0.9), but is is likely
96 to change in the future, since this may leak undesirable infor‐
97 mation into the container. If you rely on the environment being
98 available for the attached program, please use this option to be
99 future-proof. In addition to current environment variables, con‐
100 tainer=lxc will be set.
101 --clear-env
103 Clear the environment before attaching, so no undesired environ‐
104 ment variables leak into the container. The variable contain‐
105 er=lxc will be the only environment with which the attached pro‐
106 gram starts.
107 -v, --set-var variable
109 Set an additional environment variable that is seen by the at‐
110 tached program in the container. It is specified in the form of
111 "VAR=VALUE", and can be specified multiple times.
112 --keep-var variable
114 Keep a specified environment variable. It can only be specified
115 in conjunction with --clear-env, and can be specified multiple
116 times.
117
119 These options are common to most of lxc commands.
120 -?, -h, --help
122 Print a longer usage message than normal.
123 --usage
125 Give the usage message
126 -q, --quiet
128 mute on
129 -P, --lxcpath=PATH
131 Use an alternate container path. The default is /var/lib/lxc.
132 -o, --logfile=FILE
134 Output to an alternate log FILE. The default is no log.
135 -l, --logpriority=LEVEL
137 Set log priority to LEVEL. The default log priority is ERROR.
138 Possible values are : FATAL, CRIT, WARN, ERROR, NOTICE, INFO,
139 DEBUG.
140 Note that this option is setting the priority of the events log
142 in the alternate log file. It do not have effect on the ERROR
143 events log on stderr.
144 -n, --name=NAME
146 Use container identifier NAME. The container identifier format
147 is an alphanumeric string.
148 --rcfile=FILE
150 Specify the configuration file to configure the virtualization
151 and isolation functionalities for the container.
152 This configuration file if present will be used even if there is
154 already a configuration file present in the previously created
155 container (via lxc-create).
156 --version
158 Show the version number.
159
161 To spawn a new shell running inside an existing container, use
162 lxc-attach -n container
164 To restart the cron service of a running Debian container, use
167 lxc-attach -n container -- /etc/init.d/cron restart
169 To deactivate the network link eth1 of a running container that does
172 not have the NET_ADMIN capability, use either the -e option to use in‐
173 creased capabilities, assuming the ip tool is installed:
174 lxc-attach -n container -e -- /sbin/ip link delete eth1
176 Or, alternatively, use the -s to use the tools installed on the host
179 outside the container:
180 lxc-attach -n container -s NETWORK -- /sbin/ip link delete eth1
182
185 Attaching completely (including the pid and mount namespaces) to a con‐
186 tainer requires a kernel of version 3.8 or higher, or a patched kernel,
187 please see the lxc website for details. lxc-attach will fail in that
188 case if used with an unpatched kernel of version 3.7 and prior.
189 Nevertheless, it will succeed on an unpatched kernel of version 3.0 or
191 higher if the -s option is used to restrict the namespaces that the
192 process is to be attached to to one or more of NETWORK, IPC and UT‐
193 SNAME.
194 Attaching to user namespaces is supported by kernel 3.8 or higher with
196 enabling user namespace.
197
199 The Linux /proc and /sys filesystems contain information about some
200 quantities that are affected by namespaces, such as the directories
201 named after process ids in /proc or the network interface information
202 in /sys/class/net. The namespace of the process mounting the pseudo-
203 filesystems determines what information is shown, not the namespace of
204 the process accessing /proc or /sys.
205 If one uses the -s option to only attach to the pid namespace of a con‐
207 tainer, but not its mount namespace (which will contain the /proc of
208 the container and not the host), the contents of /proc will reflect
209 that of the host and not the container. Analogously, the same issue oc‐
210 curs when reading the contents of /sys/class/net and attaching to just
211 the network namespace.
212 To work around this problem, the -R flag provides the option to remount
214 /proc and /sys in order for them to reflect the network/pid namespace
215 context of the attached process. In order not to interfere with the
216 host's actual filesystem, the mount namespace will be unshared (like
217 lxc-unshare does) before this is done, essentially giving the process a
218 new mount namespace, which is identical to the hosts's mount namespace
219 except for the /proc and /sys filesystems.
220 Previous versions of lxc-attach suffered a bug whereby a user could at‐
222 tach to a containers namespace without being placed in a writeable
223 cgroup for some critical subsystems. Newer versions of lxc-attach will
224 check whether a user is in a writeable cgroup for those critical sub‐
225 systems. lxc-attach might thus fail unexpectedly for some users (E.g.
226 on systems where an unprivileged user is not placed in a writeable
227 cgroup in critical subsystems on login.). However, this behavior is
228 correct and more secure.
229
231 The -e and -s options should be used with care, as it may break the
232 isolation of the containers if used improperly.
233
235 lxc(7), lxc-create(1), lxc-copy(1), lxc-destroy(1), lxc-start(1), lxc-
236 stop(1), lxc-execute(1), lxc-console(1), lxc-monitor(1), lxc-wait(1),
237 lxc-cgroup(1), lxc-ls(1), lxc-info(1), lxc-freeze(1), lxc-unfreeze(1),
238 lxc-attach(1), lxc.conf(5)
239
241 Daniel Lezcano <daniel.lezcano@free.fr>
242 2019-09-09 lxc-attach(1)