c117ccdc5a
Some applications want to write to /proc. For instance:
docker run -it centos groupadd foo
Gives: groupadd: failure while writing changes to /etc/group
And strace reveals why:
open("/proc/self/task/13/attr/fscreate", O_RDWR) = -1 EROFS (Read-only file system)
I've looked at what other systems do, and systemd-nspawn makes /proc read-write
and /proc/sys readonly, while lxc allows "proc:mixed" which does the same,
plus it makes /proc/sysrq-trigger also readonly.
The later seems like a prudent idea, so we follows lxc proc:mixed.
Additionally we make /proc/irq and /proc/bus, as these seem to let
you control various hardware things.
Docker-DCO-1.1-Signed-off-by: Alexander Larsson <alexl@redhat.com> (github: alexlarsson)
Upstream-commit: 68493e2f7f9cb8303302e1098e3293b521ace243
Component: engine
libcontainer - reference implementation for containers
background
libcontainer specifies configuration options for what a container is. It provides a native Go implementation for using Linux namespaces with no external dependencies. libcontainer provides many convenience functions for working with namespaces, networking, and management.
container
A container is a self contained directory that is able to run one or more processes without
affecting the host system. The directory is usually a full system tree. Inside the directory
a container.json file is placed with the runtime configuration for how the processes
should be contained and ran. Environment, networking, and different capabilities for the
process are specified in this file. The configuration is used for each process executed inside the container.
See the container.json file for what the configuration should look like.
Using this configuration and the current directory holding the rootfs for a process, one can use libcontainer to exec the container. Running the life of the namespace, a pid file
is written to the current directory with the pid of the namespaced process to the external world. A client can use this pid to wait, kill, or perform other operation with the container. If a user tries to run a new process inside an existing container with a live namespace, the namespace will be joined by the new process.
You may also specify an alternate root place where the container.json file is read and where the pid file will be saved.
nsinit
nsinit is a cli application used as the reference implementation of libcontainer. It is able to
spawn or join new containers giving the current directory. To use nsinit cd into a Linux
rootfs and copy a container.json file into the directory with your specified configuration.
To execute /bin/bash in the current directory as a container just run:
nsinit exec /bin/bash
If you wish to spawn another process inside the container while your current bash session is running just run the exact same command again to get another bash shell or change the command. If the original process dies, PID 1, all other processes spawned inside the container will also be killed and the namespace will be removed.
You can identify if a process is running in a container by looking to see if pid is in the root of the directory.